Summary: | Support for real session save/restore on Wayland | ||
---|---|---|---|
Product: | [Plasma] plasmashell | Reporter: | Patrick O'Callaghan <pocallaghan> |
Component: | Session Management | Assignee: | Plasma Bugs List <plasma-bugs> |
Status: | CLOSED FIXED | ||
Severity: | task | CC: | 4wy78uwh, agurenko, akontsevich, alex, andihartmann, andreas.davour, anssi.hannula, anton.schenker, arisu+kdebugs, aspotashev, auxsvr, bdk, bednarczyk.pawel, bela73, bernie, bruno, bugs.kde.org, bugs.kde, butirsky, bvbfan, cdennett, cengique, chris, christian.rohmann, coque.couto, cribari, cruzki123, david.narvaez, dennis, desw, dkxls23, drew.m.fisher, edtoml, endymion+kde, equeim, etaash.mathamsetty, eugene.savitsky, eugene.shalygin+bugzilla.kde, feus73, forums, freisim93, funtoos, gimmemahlulz, gordan.kresic, gzatko, hoehnp, hoperidesalone, imaginator, itsforspam99, jakeskelton11, JanNowak94, jaro.schmidt, jason.boissiere, jhs, jlp, jmtornetta, johannes.hirte, john.ettedgui, julien.dlq, kairo, katze_942, kde.bugzilla.2012, kde, KDE, kde, kimiblock, kishore96, kmg952, L.Bonnaud, lassi.vaatamoinen, linux, m.kurz, madness742, mail, majzoube, matt, matteo.mazzarelli, mauromol, miranda, mmkthecoolest, mmtsales, msdobrescu, mwc85.23yp, nate, nerumo, nico.kruber, niklas, nkwkelvin, null, ol+kde, oleg.goro, omosnacek, peter, pfeiffer, pietromrl0, piotr.mierzwinski, plasma-bugs, pmrpla+bugskde, postix, rcrit, rdieter, roman, s2, sam, sandysareon, scottn, shai, shtetldik, silvan.calarco, silver.salonen, smit17xp, stefan, steve_garnier, t.p.northover, tbondvagyok, ticonzero2010, tombrown9501, victualsquid, wdmlist, whyhow+tech, winter, xqgrhiovxf6bw12h |
Priority: | HI | Keywords: | wayland-only |
Version First Reported In: | 5.24.4 | ||
Target Milestone: | 1.0 | ||
Platform: | Fedora RPMs | ||
OS: | Linux | ||
URL: | https://invent.kde.org/plasma/kwin/-/issues/113#note_500909 | ||
See Also: |
https://bugs.kde.org/show_bug.cgi?id=476831 https://bugs.kde.org/show_bug.cgi?id=15329 https://bugs.kde.org/show_bug.cgi?id=482816 https://bugs.kde.org/show_bug.cgi?id=498249 |
||
Latest Commit: | https://invent.kde.org/plasma/kwin/-/commit/54b3364a5ef7856788d509b1883123a3cd44a794 | Version Fixed In: | Plasma 6.4 or later with Qt 6.10 or later |
Sentry Crash Report: | |||
Attachments: |
attachment-25640-0.html
attachment-30227-0.html attachment-30617-0.html attachment-13297-0.html attachment-19101-0.html attachment-23823-0.html attachment-2084366-0.html attachment-3317129-0.html |
Description
Patrick O'Callaghan
2021-04-28 16:11:50 UTC
Already noted on https://community.kde.org/Plasma/Wayland_Showstoppers. We can use this as the bug report to track it. *** Bug 442192 has been marked as a duplicate of this bug. *** *** Bug 442283 has been marked as a duplicate of this bug. *** @Nate: is this going to be taken up any time soon? This is one of the biggest roadblocks to adopting plasma wayland for me. For example, its painful to rearrange firefox windows into different desktops after I start it. I have gone back to X11 solely because of this. I don't know, sorry. *** Bug 453849 has been marked as a duplicate of this bug. *** unfortuantely still an issue on plasma 5.24.90 *** Bug 455105 has been marked as a duplicate of this bug. *** *** Bug 455900 has been marked as a duplicate of this bug. *** So something like this will eventually be able to help it? https://www.youtube.com/watch?v=fRdnRwPBFBk Will this bug get fixed soon? It is the sole reason why I'm not on Wayland. Can someone let us know what time frame can be realistic expected for this to be sorted out? Thank you. It's listed on https://community.kde.org/Plasma/Wayland_Showstoppers, so it will eventually get fixed/implemented. This annoying issue is still present on Fedora 36 using KDE 5.97.0 with Wayland. Is somebody looking into this, or is there a timeline for when this will be fixed? To be honest it stops me before using/testing Plasma session with Wayland. IMO. This is one from the biggest advantage in KDE (recently even Windows 10 got it, at least for me. I don't understand why this is so low on the priority list (issue reported over 1 year ago). Would you like to contribute technical work towards fixing this issue? That's what there's not enough of right now. (In reply to Nate Graham from comment #15) > Would you like to contribute technical work towards fixing this issue? > That's what there's not enough of right now. I'm not sure if this is the question to me, because you didn't quote my statement only answered just after me. Assuming you asked me, I can reply only that maybe I would like, but technically this would be not feasible due to few free time (private reasons). Only from time to time I can report some issue and this is enough what I can do, sorry. Recently I use Neon unstable, so I think, I'm up to date with what's new in Plasma so in this way I test. Sorry for my previous complains. Hopefully this has contributed to answering your question: everyone with the technical ability to work on this issue is busy with other things that they have prioritized higher. :) (In reply to Nate Graham from comment #17) > Hopefully this has contributed to answering your question: everyone with the > technical ability to work on this issue is busy with other things that they > have prioritized higher. :) NOP for me. I work with X11, because here more things works better, or just work. Only, IMHO, KDE/Plasma a bit suffer because has less testers, because of this. And please don't understand me wrong, I would not like to set development priorities of KDE/Plasma. In general, Plasma has enough testers and QA people. What it lacks is enough developers to fix all the issues those people find. (In reply to Nate Graham from comment #19) > In general, Plasma has enough testers and QA people. What it lacks is enough > developers to fix all the issues those people find. Or the devs you already have are interested only in next shiny thing than to fix bugs in their code...:-) Sure, we can't force devs to stick around forever and fix bugs in their code but sometimes, its good to stop new development and just fix the bugs. You will have more adoption and participation, and that can potentially get you more devs. (In reply to devsk from comment #20) > Or the devs you already have are interested only in next shiny thing than to > fix bugs in their code...:-) Sure, we can't force devs to stick around > forever and fix bugs in their code but sometimes, its good to stop new > development and just fix the bugs. You will have more adoption and > participation, and that can potentially get you more devs. As a user, I also wish session restore to worked on Wayland, but inflammatory comments won't help expedite it. I would like to know what's making this difficult. X11 had a session management protocol; does Wayland define something equivalent? If not, could Plasma do something limited to KDE apps? Where is the current session save/restore code? Any documentation we could link in this bug to support someone who wants to give a shot at it? (In reply to Bernie Innocenti from comment #21) > (In reply to devsk from comment #20) > > Or the devs you already have are interested only in next shiny thing than to > > fix bugs in their code...:-) Sure, we can't force devs to stick around > > forever and fix bugs in their code but sometimes, its good to stop new > > development and just fix the bugs. You will have more adoption and > > participation, and that can potentially get you more devs. > > As a user, I also wish session restore to worked on Wayland, but > inflammatory comments won't help expedite it. I swear inflammation was not the intention. But truth can be hard sometimes and it lands wrongly in written communications. My apologies if it sounded like that. It is true that devs like to move onto newer things. I see it everyday at work. And good teams and leaders have to bring them back to the table. We have burnt sprints doing just bug fixes solely because of this. And it works! The overall quality of the code improves and customers are happier and they spread the good word! For me, with x11 applications running under x11 ad Wayland, it just seems to be a app startup mechanism. xcalc for example doesn't remain state after logging out. The observed result has been found because Firefox has a option to reopen the pages that were closed during the shutdown of Firefox. Therefore i think changing restore to reopen would be more accurate and a more concrete path to a solution. after the user ends session by logging out, rebooting, etc. Before closing the applications, the paths of the open applications should be saved to a (session) file from where these can be opened. I would be interested to know how the x11 application reopen is done in Wayland as hopefully some of the solution can be reused or be expanded on to include Wayland applications. Im not completely confident in my c++ skills yet as im studying it, so maybe some would like to help me tackle by introducing me to the project structure and the working parts making up KDE. (In reply to devsk from comment #22) > (In reply to Bernie Innocenti from comment #21) > > (In reply to devsk from comment #20) > > > Or the devs you already have are interested only in next shiny thing than to > > > fix bugs in their code...:-) Sure, we can't force devs to stick around > > > forever and fix bugs in their code but sometimes, its good to stop new > > > development and just fix the bugs. You will have more adoption and > > > participation, and that can potentially get you more devs. > > > > As a user, I also wish session restore to worked on Wayland, but > > inflammatory comments won't help expedite it. > > I swear inflammation was not the intention. But truth can be hard sometimes > and it lands wrongly in written communications. My apologies if it sounded > like that. Your comment was not inflammatory at all. It was your opinion expressed in a friendly and factual manner. IMO. But it seems to me that esp. in the FOSS community quite a few people simply can't stand critigue, however constructive or well intentioned or politely formulated it may be. They seem to experience critigue as an aggression and not as a chance for improvement. And I consider that a major weakness and an obstacle to success. This "bug" or missing key feature, as I see it, is obviously really bad and at this point of time quite shocking. Two days ago I was about switching to Plasma-Wayland because I had such a favorable impression of it. Until I discovered this thing. I don't know whether this situation is due to a lack of provision in the Wayland-design or an inability of developers to focus on what really matters. But if the Plasma devs could solve this on their own, I'd strongly suggest they drop any other Wayland related work *NOW* and start working like hell on resolving this issue. Because this is so detrimental to Plasma's and KDE's reputation that they may not recover from it. IMO. *** Bug 460567 has been marked as a duplicate of this bug. *** Same problem here with leap 15.4. Would be really nice to get it fixed ... . Hi all, same problem here does anyone know if the bug has been fixed? thanks Andrea (In reply to ticonzero from comment #28) > Hi all, > > same problem here does anyone know if the bug has been fixed? > If it had been fixed, the status would have changed. With the "Importance" downgraded to "normal", I don't expect it to be fixed before Plasma 6. ok, thanks for the clarification, I will be moving back to Xubuntu 20.04 LTS, since I make big use of this feature (In reply to ticonzero from comment #30) > ok, thanks for the clarification, I will be moving back to Xubuntu 20.04 > LTS, since I make big use of this feature I'd say that 99 % of the users make use of this feature. The rest either don't know that it exists or never log out because they are incurably addicted to Plasma-Wayland. ;) I don't think there are so many Wayland users out there because Wayland has so many restrictions and bugs so far (and I don't see any solution). Created attachment 154513 [details] attachment-25640-0.html Andreas, I would respectfully disagree. I switched to wayland once multi monitor support started working and have not looked back. Here, at least, the color is more vivid using wayland (eg X looks washed out). Aside from a few irritating bugs like this one, it works well. On Fri, Dec 9, 2022 at 12:56 PM Andreas Hartmann <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #32 from Andreas Hartmann <andihartmann@freenet.de> --- > I don't think there are so many Wayland users out there because Wayland > has so > many restrictions and bugs so far (and I don't see any solution). > > -- > You are receiving this mail because: > You are on the CC list for the bug. (In reply to Andreas Hartmann from comment #32) > I don't think there are so many Wayland users out there because Wayland has > so many restrictions and bugs so far (and I don't see any solution). Just as a reference point, Wayland session usage is shown gradually growing here: https://www.gamingonlinux.com/index.php?module=statistics&view=trends#SessionType-top Created attachment 154515 [details] attachment-30227-0.html Well I do like it, in fact I'm sticking with it, but I do miss being able to restore the old session, for me it means not having to reopen dozen of pdfs in different okular instances on different desktops On Sun, 11 Dec 2022, 20:54 Ed Tomlinson, <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #33 from Ed Tomlinson <edtoml@gmail.com> --- > Andreas, I would respectfully disagree. I switched to wayland once multi > monitor support started working and have not looked back. Here, at least, > the color is more vivid using wayland (eg X looks washed out). Aside from > a few irritating bugs like this one, it works well. > > On Fri, Dec 9, 2022 at 12:56 PM Andreas Hartmann <bugzilla_noreply@kde.org > > > wrote: > > > https://bugs.kde.org/show_bug.cgi?id=436318 > > > > --- Comment #32 from Andreas Hartmann <andihartmann@freenet.de> --- > > I don't think there are so many Wayland users out there because Wayland > > has so > > many restrictions and bugs so far (and I don't see any solution). > > > > -- > > You are receiving this mail because: > > You are on the CC list for the bug. > > -- > You are receiving this mail because: > You are on the CC list for the bug. Created attachment 154516 [details] attachment-30617-0.html You were not replying to me I guess ... On Sun, 11 Dec 2022, 21:19 ti conzero, <ticonzero2010@gmail.com> wrote: > Well I do like it, in fact I'm sticking with it, but I do miss being able > to restore the old session, for me it means not having to reopen dozen of > pdfs in different okular instances on different desktops > > On Sun, 11 Dec 2022, 20:54 Ed Tomlinson, <bugzilla_noreply@kde.org> wrote: > >> https://bugs.kde.org/show_bug.cgi?id=436318 >> >> --- Comment #33 from Ed Tomlinson <edtoml@gmail.com> --- >> Andreas, I would respectfully disagree. I switched to wayland once multi >> monitor support started working and have not looked back. Here, at least, >> the color is more vivid using wayland (eg X looks washed out). Aside from >> a few irritating bugs like this one, it works well. >> >> On Fri, Dec 9, 2022 at 12:56 PM Andreas Hartmann < >> bugzilla_noreply@kde.org> >> wrote: >> >> > https://bugs.kde.org/show_bug.cgi?id=436318 >> > >> > --- Comment #32 from Andreas Hartmann <andihartmann@freenet.de> --- >> > I don't think there are so many Wayland users out there because Wayland >> > has so >> > many restrictions and bugs so far (and I don't see any solution). >> > >> > -- >> > You are receiving this mail because: >> > You are on the CC list for the bug. >> >> -- >> You are receiving this mail because: >> You are on the CC list for the bug. > > (In reply to ticonzero from comment #35) > Well I do like it, in fact I'm sticking with it, but I do miss being able > to restore the old session, for me it means not having to reopen dozen of > pdfs in different okular instances on different desktops Same here. This bug is the only reason why I am not running Wayland. I usually have about a dozen PDF (okular) and over twenty text (kate) files open. It is too much work to open each of these files again every time I restart the computer. (In reply to Francisco Cribari from comment #37) > It is too much work to open each of these files again every time I > restart the computer. Ever heard about sleep and hibernation? ;-) (In reply to Mauro Molinari from comment #38) > Ever heard about sleep and hibernation? ;-) Sure. I use them daily. I'm on Arch Linux which is not shy in updates. Several updates (kernel (zen), etc.) require a reboot. (In reply to Francisco Cribari from comment #37) > (In reply to ticonzero from comment #35) > > > Well I do like it, in fact I'm sticking with it, but I do miss being able > > to restore the old session, for me it means not having to reopen dozen of > > pdfs in different okular instances on different desktops > > Same here. This bug is the only reason why I am not running Wayland. I > usually have about a dozen PDF (okular) and over twenty text (kate) files > open. It is too much work to open each of these files again every time I > restart the computer. If your windows don't change too much you can find a workaround on my dup: https://bugs.kde.org/show_bug.cgi?id=455900 Created attachment 154521 [details] attachment-13297-0.html It wouldn't work for me, I need to switch between different OSs and need to restart On Sun, 11 Dec 2022, 22:21 Mauro Molinari, <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #38 from Mauro Molinari <mauromol@tiscali.it> --- > (In reply to Francisco Cribari from comment #37) > > It is too much work to open each of these files again every time I > > restart the computer. > > Ever heard about sleep and hibernation? ;-) > > -- > You are receiving this mail because: > You are on the CC list for the bug. (In reply to John E from comment #40) > (In reply to Francisco Cribari from comment #37) > > (In reply to ticonzero from comment #35) > > > > > Well I do like it, in fact I'm sticking with it, but I do miss being able > > > to restore the old session, for me it means not having to reopen dozen of > > > pdfs in different okular instances on different desktops > > > > Same here. This bug is the only reason why I am not running Wayland. I > > usually have about a dozen PDF (okular) and over twenty text (kate) files > > open. It is too much work to open each of these files again every time I > > restart the computer. > > If your windows don't change too much you can find a workaround on my dup: > https://bugs.kde.org/show_bug.cgi?id=455900 In my view, the most obvious and promising "workaround" for an ordinary user is to continue enjoying Plasma-X11 until all major issues in Plasma-Wayland are fixed and it's ready for productive use. ;) Created attachment 154527 [details] attachment-19101-0.html I installed kubuntu 22.04LTS, that came with Wayland, how can I get back to X11-Plasma, without returning to 18.04LTS ? On Mon, 12 Dec 2022, 10:10 , <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #42 from imaginator@mailbox.org --- > (In reply to John E from comment #40) > > (In reply to Francisco Cribari from comment #37) > > > (In reply to ticonzero from comment #35) > > > > > > > Well I do like it, in fact I'm sticking with it, but I do miss being > able > > > > to restore the old session, for me it means not having to reopen > dozen of > > > > pdfs in different okular instances on different desktops > > > > > > Same here. This bug is the only reason why I am not running Wayland. I > > > usually have about a dozen PDF (okular) and over twenty text (kate) > files > > > open. It is too much work to open each of these files again every time > I > > > restart the computer. > > > > If your windows don't change too much you can find a workaround on my > dup: > > https://bugs.kde.org/show_bug.cgi?id=455900 > > In my view, the most obvious and promising "workaround" for an ordinary > user is > to continue enjoying Plasma-X11 until all major issues in Plasma-Wayland > are > fixed and it's ready for productive use. ;) > > -- > You are receiving this mail because: > You are on the CC list for the bug. (In reply to ticonzero from comment #43) > Created attachment 154527 [details] > attachment-19101-0.html > > I installed kubuntu 22.04LTS, that came with Wayland, how can I get back to > X11-Plasma, without returning to 18.04LTS ? I think that's a question for the kbuntu support forum / mailing list. Created attachment 154528 [details] attachment-23823-0.html Thanks! On Mon, 12 Dec 2022, 10:41 , <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #44 from imaginator@mailbox.org --- > (In reply to ticonzero from comment #43) > > Created attachment 154527 [details] > > attachment-19101-0.html > > > > I installed kubuntu 22.04LTS, that came with Wayland, how can I get back > to > > X11-Plasma, without returning to 18.04LTS ? > I think that's a question for the kbuntu support forum / mailing list. > > -- > You are receiving this mail because: > You are on the CC list for the bug. (In reply to imaginator from comment #42) > (In reply to John E from comment #40) > > (In reply to Francisco Cribari from comment #37) > > > (In reply to ticonzero from comment #35) > > > > > > > Well I do like it, in fact I'm sticking with it, but I do miss being able > > > > to restore the old session, for me it means not having to reopen dozen of > > > > pdfs in different okular instances on different desktops > > > > > > Same here. This bug is the only reason why I am not running Wayland. I > > > usually have about a dozen PDF (okular) and over twenty text (kate) files > > > open. It is too much work to open each of these files again every time I > > > restart the computer. > > > > If your windows don't change too much you can find a workaround on my dup: > > https://bugs.kde.org/show_bug.cgi?id=455900 > > In my view, the most obvious and promising "workaround" for an ordinary user > is to continue enjoying Plasma-X11 until all major issues in Plasma-Wayland > are fixed and it's ready for productive use. ;) That's what I do as well, but it's not really what I would call a workaround. Exactly same problem can be seen even if KDE is run in wayland mode ... . Please ignore my last entry - wrong ticket :-). I've been "forced" to move now to wayland to get the per screen scale factor as it will never be implemented on X11. But loosing restore session is a pain. It seems we miss a global review on plasma development, with clear objectives for each release, to make all users (and developers) happy. (In reply to Bruno Friedmann from comment #49) > I've been "forced" to move now to wayland to get the per screen scale factor > as it will never be implemented on X11. But loosing restore session is a > pain. > It seems we miss a global review on plasma development, with clear > objectives for each release, to make all users (and developers) happy. I'm sure it holds the top position on their priority list for Plasma 6. But: who knows? ;) (In reply to Bernie Innocenti from comment #21) > I would like to know what's making this difficult. X11 had a session > management protocol; does Wayland define something equivalent? Let me loosely quote the wiki [1] > Session restoring does not include Wayland native windows (...). > Our sessions management recover engine is based on the X Session Management Protocol > and there is apparently currently no generic concept to do it on Wayland. > But on Qt it's plugin-able and GNOME has had their own implementation for some time. [2] Regarding opened discussions, there's an issue in the kwin repo, see "Goal 2: Session restore" [3] [1] https://community.kde.org/Plasma/Wayland_Showstoppers [2] https://wiki.gnome.org/Projects/SessionManagement/GnomeSession [3] https://invent.kde.org/plasma/kwin/-/issues/113#note_500909 See also the MR linked in the issue above: "Draft: wayland: Add support for xdg-session-v1" [1] [1] https://invent.kde.org/plasma/kwin/-/merge_requests/3024 Looks like some work was started 3 months ago and then, it stopped. I tried to use wayland again to see if the experience is any better and went back to x11 after an hour of frustration. I have 4 different desktops for various projects and just restoring all my konsole windows (about 9 of them) and their tabs, and placing them in their respective places on each boot is a big pain. Let's not even talk about putting other apps where I am used to seeing them. How does a serious dev use wayland at all? I am just curious. (In reply to devsk from comment #53) > Looks like some work was started 3 months ago and then, it stopped. > > I tried to use wayland again to see if the experience is any better and went > back to x11 after an hour of frustration. I have 4 different desktops for > various projects and just restoring all my konsole windows (about 9 of them) > and their tabs, and placing them in their respective places on each boot is > a big pain. Let's not even talk about putting other apps where I am used to > seeing them. > > How does a serious dev use wayland at all? I am just curious. My thoughts exactly. I've no doubt that serious devs *do* use Wayland, but I've no idea what their workflow is like. For me, session restore is an absolute requirement, so I keep feeling I must be missing something. (In reply to Patrick O'Callaghan from comment #54) > (In reply to devsk from comment #53) > > Looks like some work was started 3 months ago and then, it stopped. > > > > I tried to use wayland again to see if the experience is any better and went > > back to x11 after an hour of frustration. I have 4 different desktops for > > various projects and just restoring all my konsole windows (about 9 of them) > > and their tabs, and placing them in their respective places on each boot is > > a big pain. Let's not even talk about putting other apps where I am used to > > seeing them. > > > > How does a serious dev use wayland at all? I am just curious. > > My thoughts exactly. I've no doubt that serious devs *do* use Wayland, but > I've no idea what their workflow is like. For me, session restore is an > absolute requirement, so I keep feeling I must be missing something. Session restore is such a basic feature nowadays that it seems crazy to even have to waste one thought on it. It *has* to be there. Like wheels on a car. If not, something went seriously wrong somewhere. My way to deal with that - IMO disappointing - Wayland-show is that I simply don't care any more. To me it doesn't offer any benefit. And as long as there's a Plasma-X11 I'm happy. Are there any reasonable workarounds? Say a script that reads all window positions, sizes, etc., and saves the information, and then another script that can read this info and open and arrange the windows? For a while I used a script based on: https://github.com/zepalmer/script-vdr but it's for Python-2 and X11 and I haven't tried to bring it up to date. It's based around wmctrl, which AFAIK is not supported on Wayland, but there might be something useful there. Note that it's designed to save/restore the windows belonging to one process, e.g. your browser, not the full user session. Still an issue on Kubuntu 22.04, going to switch back to X until it is fixed. According to the plasma/wayland showstoppers page, session restore is a priority: https://community.kde.org/Plasma/Wayland_Showstoppers However, on the plasma 6 development page there is no mention of session restore: https://community.kde.org/Plasma/Plasma_6 Does anyone know if session restore is supposed to be in plasma 6? > Does anyone know if session restore is supposed to be in plasma 6? The plan is to solve all Wayland Showstoppers - including session restore - until Plasma 6.0 is released [1] [1] https://pointieststick.com/2023/09/06/september-plasma-6-update/#comment-37811 from my understanding the issue was downgraded to a non-showstopper anymore. At least this is shown now in the wiki page (https://community.kde.org/Plasma/Wayland_Showstoppers). I hope that Qt-pluginable option will still be provided on plasma 6. Is there any update about when it will be integrated? The plan right now is as follows: - For Plasma 6.0, implement a sort of fake session restore that on login, simply re-opens apps that were open on last logout, and count on apps that have state worth saving doing the saving themselves (and many apps already do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). - For a later Plasma 6 version, implement real session restore that's governed by a new Wayland protocol (one is in progress) and make KDE apps opt into it. - After that, keep the fake session restore and use it for apps that don't opt into the real session restore. This should ultimately produce a better UX than on X11, where many apps never opted into session restore and so it was semi-random as to which apps got launched on login. Will what is planned for KDE 6 take into account multiple monitors and multiple virtual desktops so that apps start in the desired monitor/virtual desktop? That is crucial! On 11/11/23 17:38, Charles Dennett wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #63 from Charles Dennett <cdennett@gmail.com> --- > Will what is planned for KDE 6 take into account multiple monitors and multiple > virtual desktops so that apps start in the desired monitor/virtual desktop? > For the fake session restore, I can't make any promises about that. I've talked to the fellow doing the work and he wants to take those into consideration, but it becomes more technically challenging, so the first version might not. But I do expect those to *eventually* work in the fake session restore. (In reply to Charles Dennett from comment #63) > Will what is planned for KDE 6 take into account multiple monitors and > multiple virtual desktops so that apps start in the desired monitor/virtual > desktop? And please don't forget about activities! I'm tired of moving Firefox windows to different activities manually each time after login even on KDE 5 X11 session. (In reply to Nate Graham from comment #62) > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, […] Wow, that sounds great! I mean, real session restore would be a fantastic feature to have, but the “fake” option is already a great improvement to what we have at the moment (i.e. nothing). I got already worried this was dropped entirely when I saw the change in the showstopper bug list. This is after all by far the biggest pain point I have with KDE Wayland. (In reply to Nate Graham from comment #62) > The plan right now is as follows: > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, and count on apps that > have state worth saving doing the saving themselves (and many apps already > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). Aside from the virtual desktop issue already mentioned by several people, it's worth noting that Dolphin does not currently restore its state correctly even under X11. Briefly, if you have more than one Dolphin window, some with the Terminal open and some not, the restored Dolphin will have all windows showing the Terminal in the same state, either open or closed, apparently at random. I reported this over a year ago (see https://bugs.kde.org/show_bug.cgi?id=455324) but never saw a reply. Indeed, Dolphin lost a lot of its session features, like restoring at the original position. For me, being an Nvidia user (didn't know the drawbacks at the time and Plasma looked to work good enough with x11 and Nvidia), has become more and more unlikely I would move to the next Plasma at all. This combination that makes any input unusable, the mouse pointer or the keyboard signals being sent to Plasma after 30 secs or so, but also lacking session, are unacceptable. I understand some features are ~ maybe ~ difficult to implement, but basic features like these are a must. I can't expect from a user to rather change a laptop just because he would use a DE, sorry. (In reply to Nate Graham from comment #62) > The plan right now is as follows: > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, and count on apps that > have state worth saving doing the saving themselves (and many apps already > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). > - For a later Plasma 6 version, implement real session restore that's > governed by a new Wayland protocol (one is in progress) and make KDE apps > opt into it. > - After that, keep the fake session restore and use it for apps that don't > opt into the real session restore. > > This should ultimately produce a better UX than on X11, where many apps > never opted into session restore and so it was semi-random as to which apps > got launched on login. Does Konsole save session (the windows and tabs and the current directory of each tab) by itself? My experience is that it does not, but given that things like Kate does, why can't Konsole, which is also a KDE app, do that? Am I missing something? > Does Konsole save session (the windows and tabs and the current directory of each tab) by itself? My experience is that it does not, but given that things like Kate does, why can't Konsole, which is also a KDE app, do that?
> Am I missing something?
Yes my experience proove it, I usually have 10 tabs opened in konsole, most of them are dedicated to a relative work, and as such the directory in which the last session was run, and is restored on the right desktop.
That IS the most reason I don't use (yet) Wayland.
(In reply to Bruno Friedmann from comment #71) > > Does Konsole save session (the windows and tabs and the current directory of each tab) by itself? My experience is that it does not, but given that things like Kate does, why can't Konsole, which is also a KDE app, do that? > > Am I missing something? > > Yes my experience proove it, I usually have 10 tabs opened in konsole, most > of them are dedicated to a relative work, and as such the directory in which > the last session was run, and is restored on the right desktop. > That IS the most reason I don't use (yet) Wayland. One workaround for Konsole exists on Stack-Exchange: https://unix.stackexchange.com/a/593779 It also works on Wayland. However, I believe this should be possible with built-in functionality. the same here. On 02/12/23 09:36, Bruno Friedmann wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #71 from Bruno Friedmann <bruno@ioda-net.ch> --- >> Does Konsole save session (the windows and tabs and the current directory of each tab) by itself? My experience is that it does not, but given that things like Kate does, why can't Konsole, which is also a KDE app, do that? >> Am I missing something? > Yes my experience proove it, I usually have 10 tabs opened in konsole, most of > them are dedicated to a relative work, and as such the directory in which the > last session was run, and is restored on the right desktop. > That IS the most reason I don't use (yet) Wayland. > (In reply to Nate Graham from comment #62) > The plan right now is as follows: > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, and count on apps that > have state worth saving doing the saving themselves (and many apps already > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). > - For a later Plasma 6 version, implement real session restore that's > governed by a new Wayland protocol (one is in progress) and make KDE apps > opt into it. > - After that, keep the fake session restore and use it for apps that don't > opt into the real session restore. > > This should ultimately produce a better UX than on X11, where many apps > never opted into session restore and so it was semi-random as to which apps > got launched on login. Does that plan also apply to Plasma-6-X11 or will session restore there just keep working like in Plasma-5-X11? Thanks. I have made two scripts (heavily modified from scripts that I found somewhere else) to help session restore semi-automatically: https://github.com/Kelvin-Ng/konsole-session-restore: This script saves and reopens the Konsole windows with original tabs and directories. https://github.com/Kelvin-Ng/kde-window-status-restore: This script saves and restores the statuses of the windows, currently only activity and desktop. Note that it does not reopen the windows. The windows will have to be first reopened before running the restore script. Hope these scripts are helpful. (In reply to Nate Graham from comment #62) > The plan right now is as follows: > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, and count on apps that > have state worth saving doing the saving themselves (and many apps already > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). > - For a later Plasma 6 version, implement real session restore that's > governed by a new Wayland protocol (one is in progress) and make KDE apps > opt into it. > - After that, keep the fake session restore and use it for apps that don't > opt into the real session restore. > > This should ultimately produce a better UX than on X11, where many apps > never opted into session restore and so it was semi-random as to which apps > got launched on login. Alright, maybe my question was a bit too difficult to understand. So I rephrase it a bit: will session restore in Plasma-6-X11 work _right from the start_ as it does in Plasma-5-X11 or not? Some people don't give a damn about Wayland but do care about reliable and complete session restore. Created attachment 164287 [details] attachment-2084366-0.html Count me among those who need reliable session restore, I'm stuck with xubuntu 20.4 and will stay there until session restore will work properly On Tue, 19 Dec 2023, 10:07 , <bugzilla_noreply@kde.org> wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #76 from imaginator@mailbox.org --- > (In reply to Nate Graham from comment #62) > > The plan right now is as follows: > > - For Plasma 6.0, implement a sort of fake session restore that on login, > > simply re-opens apps that were open on last logout, and count on apps > that > > have state worth saving doing the saving themselves (and many apps > already > > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). > > - For a later Plasma 6 version, implement real session restore that's > > governed by a new Wayland protocol (one is in progress) and make KDE apps > > opt into it. > > - After that, keep the fake session restore and use it for apps that > don't > > opt into the real session restore. > > > > This should ultimately produce a better UX than on X11, where many apps > > never opted into session restore and so it was semi-random as to which > apps > > got launched on login. > > Alright, maybe my question was a bit too difficult to understand. So I > rephrase it a bit: will session restore in Plasma-6-X11 work _right from > the > start_ as it does in Plasma-5-X11 or not? Some people don't give a damn > about > Wayland but do care about reliable and complete session restore. > > -- > You are receiving this mail because: > You are on the CC list for the bug. (In reply to imaginator from comment #76) > Alright, maybe my question was a bit too difficult to understand. So I > rephrase it a bit: will session restore in Plasma-6-X11 work _right from the > start_ as it does in Plasma-5-X11 or not? Some people don't give a damn > about Wayland but do care about reliable and complete session restore. On X11? No changes have been made to session restore on X11 there so I would expect so. However I can't personally guarantee no regressions, as I'm not personally using X11 anymore and therefore not testing for regressions there. Someone else will need to step in for that. (In reply to Nate Graham from comment #78) > (In reply to imaginator from comment #76) > > Alright, maybe my question was a bit too difficult to understand. So I > > rephrase it a bit: will session restore in Plasma-6-X11 work _right from the > > start_ as it does in Plasma-5-X11 or not? Some people don't give a damn > > about Wayland but do care about reliable and complete session restore. > > On X11? No changes have been made to session restore on X11 there so I would > expect so. However I can't personally guarantee no regressions, as I'm not > personally using X11 anymore and therefore not testing for regressions > there. Someone else will need to step in for that. Thanks for clarifying this. *** Bug 476831 has been marked as a duplicate of this bug. *** *** Bug 481745 has been marked as a duplicate of this bug. *** Promoting this to be a 15-minute bug since it's Wayland-only and the Wayland-session is the default in Plasma 6 now. *** Bug 479886 has been marked as a duplicate of this bug. *** We've introduced a fallback. plasma-fallback-session-save which provides something. It's not perfect. It just sees which apps were open and then reopens them. The application has no mechanism to save any state. There's definitely room for improvement, but that involves upstream and updating all applications(!) which isn't worth us tracking in our bug tracker. Git commit 804976c5ecec1fbf5f6e7e09970a8269bdf748d2 by David Edmundson. Committed on 08/03/2024 at 09:17. Pushed by davidedmundson into branch 'master'. Migrate manual session saving to plasma-shutdown This allows the wayland fallback session saving to run too and removes more ksmserver logic from our library code. session-restore is patched to also run when the configuration file is set to restore the manually saved session. M +0 -3 libkworkspace/CMakeLists.txt M +2 -3 libkworkspace/sessionmanagement.cpp M +1 -0 startkde/plasma-shutdown/org.kde.Shutdown.xml M +24 -2 startkde/plasma-shutdown/shutdown.cpp M +1 -0 startkde/plasma-shutdown/shutdown.h M +1 -1 startkde/session-restore/restore.cpp https://invent.kde.org/plasma/plasma-workspace/-/commit/804976c5ecec1fbf5f6e7e09970a8269bdf748d2 > We've introduced a fallback. plasma-fallback-session-save which provides something. Thanks! > It's not perfect. It just sees which apps were open and then reopens them. The application has no mechanism to save any state. Let's take Konsole as an example. Does this mean that Konsole will not save its current directory? Correct. Because real session restore is not implemented on Wayland yet, at the moment this manual session restore hooks into the "real fake session restore" mechanism that will be used until we get real session restore. (In reply to Nate Graham from comment #87) > Correct. Because real session restore is not implemented on Wayland yet, at > the moment this manual session restore hooks into the "real fake session > restore" mechanism that will be used until we get real session restore. Any idea on when real session-restore might be coming? Once the proposed Wayland protocol gets approved and merged, support in Plasma will likely arrive quickly. I don't know if it's helpful to speculate as to when the protocol might get approved. It could be tomorrow and it could be in two years. Obviously we all would prefer the former over the latter. :) How about using this piece of software? https://github.com/Vladimir-csp/uwsm Session is restored in strange way. Firstly starts applications, so we have black screen and starting apps. and at end starts plasmashell. Second thing is that session in konsole is not restored, but in IceWm its working well. (In reply to Piotr Mierzwinski from comment #91) > Session is restored in strange way. Firstly starts applications, so we have > black screen and starting apps. and at end starts plasmashell. > Second thing is that session in konsole is not restored, but in IceWm its > working well. Applications also start in front of a black screen for me as well in 5.27 on X (and have in 5.24 as well). >How about using this piece of software?
That's not what's missing, that's a replacement for what we call "startkde"
(In reply to Nate Graham from comment #62) > The plan right now is as follows: > - For Plasma 6.0, implement a sort of fake session restore that on login, > simply re-opens apps that were open on last logout, and count on apps that > have state worth saving doing the saving themselves (and many apps already > do; for example Firefox, Discord, Dolphin, Kate, Kile, Elisa...). > - For a later Plasma 6 version, implement real session restore that's > governed by a new Wayland protocol (one is in progress) and make KDE apps > opt into it. > - After that, keep the fake session restore and use it for apps that don't > opt into the real session restore. > > This should ultimately produce a better UX than on X11, where many apps > never opted into session restore and so it was semi-random as to which apps > got launched on login. I've just installed Fedora 40 with Plasma 6 on Wayland. I'm sorry to say the "fake session restore" simply doesn't work. The only app that re-opens on login is (ironically) Firefox, which isn't even a Plasma app. Both Konsole and Dolphin fail to appear and have to be started manually every time. Needless to say, this is a considerable inconvenience. As you just quoted "The plan right now is as follows:" this was the plan back in 2023. If you look at the "Version Fixed in" on the top right, it says 6.1. You have to wait til June or so. :) (In reply to postix from comment #95) > As you just quoted "The plan right now is as follows:" this was the plan > back in 2023. If you look at the "Version Fixed in" on the top right, it > says 6.1. You have to wait til June or so. :) In a post from January, Neil said he hoped to have this working for F40: https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/thread/PMAYGU3UJ6RXEOBGQJD6ZPKBLXEYLNJ2/ (you'll have to scroll down. The current list archive system doesn't seem to offer links to a specific message). (In reply to Patrick O'Callaghan from comment #96) > > (you'll have to scroll down. The current list archive system doesn't seem to > offer links to a specific message). Click on the link symbol in the upper right corner of the message. https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/message/HP26SLPOXK264324MAJMVC24MDEMOXQ7/ (In reply to Bernd Steinhauser from comment #97) > (In reply to Patrick O'Callaghan from comment #96) > > > > (you'll have to scroll down. The current list archive system doesn't seem to > > offer links to a specific message). > > Click on the link symbol in the upper right corner of the message. > > https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/ > message/HP26SLPOXK264324MAJMVC24MDEMOXQ7/ Thanks. The reference is: https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/message/XSBZENY3VQCY57W3GXGCL7RMFZDANLTO/ This is marked RESOLVED as FIXED for a while now. Do people have experience with the new session restore in Plasma 6? How good is it working with X11 and wayland? No signs of session restore as of 6.0.90 The hack is definitely there. It follows the systemsettings -> desktop session -> session restore settings It is set to "On last logout" here, but nothing gets launched on the next log in. Same here using Gentoo. Oh, I'm on Gentoo as well. Maybe it is our packaging that disables this feature? David, could you, please, point out the implementation location so I can try to figure out why it does not work? Could be related, could be not but I am "loosing" configurations set in system settings since a couple of versions. Not sure when this starts to happen but sure after the release of the first "stable" version. For example, I am lousing manual file association that I have configured, the landing page of system settings is no longer showing me the last used configurations pages and the restored session is not working. These are the ones I have noticed. But the theme is correctly saved and restored, for example. Could be that these configs are stored in places that gentoo eats or has set the wrong permissions? Any log file I could check for this kind of errors? The key point is that I have recently switched to wayland. The FIRST log in, these things were working (so I have correct file associations, a correct restored session and the used last used config pages) but as soon as I reboot / log off (not sure what was the action), I lost all of them. In any case, the problem with file association and landing page of system settings is also happening on xorg (but not the restored session, that is working fine on xorg). Should I file a different bug report or do you think it could be the same problem? Should I fill this in gentoo bugtracker or here? (In reply to Cruz Enrique from comment #105) > Could be related, could be not but I am "loosing" configurations set in > system settings since a couple of versions. Not sure when this starts to > happen but sure after the release of the first "stable" version. For > example, I am lousing manual file association that I have configured, the > landing page of system settings is no longer showing me the last used > configurations pages and the restored session is not working. These are the > ones I have noticed. But the theme is correctly saved and restored, for > example. Could be that these configs are stored in places that gentoo eats > or has set the wrong permissions? Any log file I could check for this kind > of errors? > > The key point is that I have recently switched to wayland. The FIRST log in, > these things were working (so I have correct file associations, a correct > restored session and the used last used config pages) but as soon as I > reboot / log off (not sure what was the action), I lost all of them. In any > case, the problem with file association and landing page of system settings > is also happening on xorg (but not the restored session, that is working > fine on xorg). > > Should I file a different bug report or do you think it could be the same > problem? Should I fill this in gentoo bugtracker or here? Please file a separate bug report. I don't think this is related. (By the way, the word you're looking for is "losing", not "loosing". Two different things. No offence but this is a common mistake). (In reply to Cruz Enrique from comment #105) > Could be related, could be not but I am "loosing" configurations set in > system settings since a couple of versions. Not sure when this starts to > happen but sure after the release of the first "stable" version. For > example, I am lousing manual file association that I have configured, the > landing page of system settings is no longer showing me the last used > configurations pages and the restored session is not working. These are the > ones I have noticed. But the theme is correctly saved and restored, for > example. Could be that these configs are stored in places that gentoo eats > or has set the wrong permissions? Any log file I could check for this kind > of errors? > > The key point is that I have recently switched to wayland. The FIRST log in, > these things were working (so I have correct file associations, a correct > restored session and the used last used config pages) but as soon as I > reboot / log off (not sure what was the action), I lost all of them. In any > case, the problem with file association and landing page of system settings > is also happening on xorg (but not the restored session, that is working > fine on xorg). > > Should I file a different bug report or do you think it could be the same > problem? Should I fill this in gentoo bugtracker or here? Please file a separate bug report. I don't think this is related. (By the way, the word you're looking for is "losing", not "loosing". Two different things. No offence but this is a common mistake).(In reply to devsk from comment #99) > This is marked RESOLVED as FIXED for a while now. > > Do people have experience with the new session restore in Plasma 6? How good > is it working with X11 and wayland? Not as yet on Fedora 40, Plasma 6.0.5. I think 6.1 is expected soon and hopefully Fake Session Restore will be part of it. The fake session restore is part of plasma 6.1 and not available in 6.0 as far as I can tell. However, as stated in the link in Comment 98, Fedora has backported the session restore to their plasma packages (version 6.0) for Fedora 40. Hope this clears up that confusion. Here the link to the commit for the Fedora 40 package: https://src.fedoraproject.org/rpms/plasma-workspace/c/d03958da910f3ba08c531e7569ff40c19ca2f045?branch=f40 (In reply to Armin from comment #109) > Here the link to the commit for the Fedora 40 package: > https://src.fedoraproject.org/rpms/plasma-workspace/c/ > d03958da910f3ba08c531e7569ff40c19ca2f045?branch=f40 Note that this is not yet available in the Fedora repos, so probably better to wait. Regarding the actual session restore functionality, I have just switched to Fedora 40 and tested this a bit. So far, the fake session restore was not useful at all since it only launches the applications, but those apparently are not capable of remembering their previous state. For instance, I rebooted my machine with 1 Kate, 2 Okular and 2 Konsole instances open, and while Kate, 2 Okular and 2 Konsole windows opened after the restart, none of those opened the previously open document or changed to the directory they previously were at. Back in X11, the session restore would not only reopen the applications, but also restore the actual documents (e.g. reopen the PDF file in Okular, change to the directory in Konsole, or restore a session in Kate). As such, the current session restore doesn't add any value, since I need to manually open the documents/session anyway myself. (In reply to Patrick O'Callaghan from comment #110) > (In reply to Armin from comment #109) > > Here the link to the commit for the Fedora 40 package: > > https://src.fedoraproject.org/rpms/plasma-workspace/c/ > > d03958da910f3ba08c531e7569ff40c19ca2f045?branch=f40 > > Note that this is not yet available in the Fedora repos, so probably better > to wait. Ok, what makes you think it is not? As far as I can tell, the F40 spec file contains the patch: https://src.fedoraproject.org/rpms/plasma-workspace/blob/f40/f/plasma-workspace.spec#_33 Created attachment 170201 [details] attachment-3317129-0.html here I have Kubuntu 20.04 and plasmashell 5.18.8 works (almost) perfectly: - it reopens okular, konsoles, kate, thunderbird and multiple firefoxes in the same Desktops, same location, on the same files or webpage - it reopens Tekstudio in the wrong desktop with the same files open - it reopens gvim and Paraview with no file open I'm stuck with Kubuntu 20.04 because I just need this feature On 06/06/24 12:54, Armin wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #111 from Armin<dkxls23@gmail.com> --- > Regarding the actual session restore functionality, I have just switched to > Fedora 40 and tested this a bit. So far, the fake session restore was not > useful at all since it only launches the applications, but those apparently are > not capable of remembering their previous state. > > For instance, I rebooted my machine with 1 Kate, 2 Okular and 2 Konsole > instances open, and while Kate, 2 Okular and 2 Konsole windows opened after the > restart, none of those opened the previously open document or changed to the > directory they previously were at. Back in X11, the session restore would not > only reopen the applications, but also restore the actual documents (e.g. > reopen the PDF file in Okular, change to the directory in Konsole, or restore a > session in Kate). > > As such, the current session restore doesn't add any value, since I need to > manually open the documents/session anyway myself. > (In reply to Armin from comment #112) > (In reply to Patrick O'Callaghan from comment #110) > > (In reply to Armin from comment #109) > > > Here the link to the commit for the Fedora 40 package: > > > https://src.fedoraproject.org/rpms/plasma-workspace/c/ > > > d03958da910f3ba08c531e7569ff40c19ca2f045?branch=f40 > > > > Note that this is not yet available in the Fedora repos, so probably better > > to wait. > > Ok, what makes you think it is not? As far as I can tell, the F40 spec file > contains the patch: > https://src.fedoraproject.org/rpms/plasma-workspace/blob/f40/f/plasma- > workspace.spec#_33 Because of: https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/message/HP26SLPOXK264324MAJMVC24MDEMOXQ7/ which mentions 6.1 "due out in June". I know we're in June now, but 6.1 is not out yet, at least in stable Fedora repos. (In reply to Armin from comment #111) > Regarding the actual session restore functionality, I have just switched to > Fedora 40 and tested this a bit. So far, the fake session restore was not > useful at all since it only launches the applications, but those apparently > are not capable of remembering their previous state. > > For instance, I rebooted my machine with 1 Kate, 2 Okular and 2 Konsole > instances open, and while Kate, 2 Okular and 2 Konsole windows opened after > the restart, none of those opened the previously open document or changed to > the directory they previously were at. Back in X11, the session restore > would not only reopen the applications, but also restore the actual > documents (e.g. reopen the PDF file in Okular, change to the directory in > Konsole, or restore a session in Kate). > > As such, the current session restore doesn't add any value, since I need to > manually open the documents/session anyway myself. "If things don't come easy, there is no premium on effort." (Branch Rickey). So why don't you just dump Plasma-Wayland and use Plasma-X11 until this essential (but nowadays rather trivial) feature is _fully_ implemented? At least from a productivity point of view a masochistic adherence to Wayland doesn't make sense. (In reply to Patrick O'Callaghan from comment #114) > (In reply to Armin from comment #112) > > (In reply to Patrick O'Callaghan from comment #110) > > > (In reply to Armin from comment #109) > > > > Here the link to the commit for the Fedora 40 package: > > > > https://src.fedoraproject.org/rpms/plasma-workspace/c/ > > > > d03958da910f3ba08c531e7569ff40c19ca2f045?branch=f40 > > > > > > Note that this is not yet available in the Fedora repos, so probably better > > > to wait. > > > > Ok, what makes you think it is not? As far as I can tell, the F40 spec file > > contains the patch: > > https://src.fedoraproject.org/rpms/plasma-workspace/blob/f40/f/plasma- > > workspace.spec#_33 > > Because of: > > https://lists.fedoraproject.org/archives/list/kde@lists.fedoraproject.org/ > message/HP26SLPOXK264324MAJMVC24MDEMOXQ7/ > > which mentions 6.1 "due out in June". I know we're in June now, but 6.1 is > not out yet, at least in stable Fedora repos. What's mentioned there as "rudimentary support" for 6.1 seems to be what's described in comment 84. For me that wouldn't be enough. (In reply to ticonzero from comment #113) > Created attachment 170201 [details] > attachment-3317129-0.html > > here I have Kubuntu 20.04 and plasmashell 5.18.8 works (almost) perfectly: > > - it reopens okular, konsoles, kate, thunderbird and multiple firefoxes > in the same Desktops, same location, on the same files or webpage > > - it reopens Tekstudio in the wrong desktop with the same files open > > - it reopens gvim and Paraview with no file open > > > I'm stuck with Kubuntu 20.04 because I just need this feature You're not stuck there. Session restore (still) works on the X11 session on Plasma 5.24 (Kubuntu 22.04) and Plasma 5.27 (Kubuntu 24.04). thanks, I will give it a try On 06/06/24 16:29, Niklas Sombert wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #117 from Niklas Sombert <niklas@ytvwld.de> --- > (In reply to ticonzero from comment #113) >> Created attachment 170201 [details] >> attachment-3317129-0.html >> >> here I have Kubuntu 20.04 and plasmashell 5.18.8 works (almost) perfectly: >> >> - it reopens okular, konsoles, kate, thunderbird and multiple firefoxes >> in the same Desktops, same location, on the same files or webpage >> >> - it reopens Tekstudio in the wrong desktop with the same files open >> >> - it reopens gvim and Paraview with no file open >> >> >> I'm stuck with Kubuntu 20.04 because I just need this feature > You're not stuck there. Session restore (still) works on the X11 session on > Plasma 5.24 (Kubuntu 22.04) and Plasma 5.27 (Kubuntu 24.04). > So, after reading the comments, it looks like it is still the "same old same old". Sticking with plasma 5 then! Just a heads-up to note that Plasma 6.1 is now available. I have installed it on Fedora 40 (via the normal dnf upgrade system) and look forward to testing it. Updated Arch with Plasma 6.1, rebooted, didn't work. Then, switched setting to: "When session was manually saved" and rebooted. Don't remember seeing a button to "Save Session" but it worked same as setting: "On last logout". Switched back to "On last logout" and it is still working. Very pleased! Thanks to everyone who put in the hard work to make this happen for Plasma 6. > Switched back to "On last logout" and it is still working. Very pleased! Thanks to everyone who put in the hard work to make this happen for Plasma 6.
What you are saying is that switching away and the switching back to this option made the session restore work properly for you with 6.1.
Can you please confirm what all works? Does it put all your console windows with their tabs in their respective desktops? How about firefox windows?
What else did you try to restore?
restoring okular is also important for me On 21/06/24 08:50, devsk wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #122 from devsk <funtoos@yahoo.com> --- >> Switched back to "On last logout" and it is still working. Very pleased! Thanks to everyone who put in the hard work to make this happen for Plasma 6. > What you are saying is that switching away and the switching back to this > option made the session restore work properly for you with 6.1. > > Can you please confirm what all works? Does it put all your console windows > with their tabs in their respective desktops? How about firefox windows? > > What else did you try to restore? > (In reply to Patrick O'Callaghan from comment #120) > Just a heads-up to note that Plasma 6.1 is now available. I have installed > it on Fedora 40 (via the normal dnf upgrade system) and look forward to > testing it. Here's my experience after brief testing: * Apps are restored, including non-KDE ones such as Firefox and Evolution, but Yakuake does not restore. * They all appear on the first desktop and have to be manually repositioned. Window sizes are correct but window positioning within the desktop is arbitrary. * AFAIK each app is responsible for its own internal state. Konsole does not preserve tabs. Dolphin does not preserve terminal open/closed state independently for each window (though all windows are present) and instead of 3 windows in different directories they are all copies of each other (probably a Dolphin problem). * Shortcut for normal interactive logout doesn't work consistently, not does the Logout menu selection. The shortcut for immediate logout (Shift-Ctrl-Alt-Del) does work. Login/logout is very fast compared to 6.0 on Wayland. (In reply to Patrick O'Callaghan from comment #124) > (In reply to Patrick O'Callaghan from comment #120) > > Just a heads-up to note that Plasma 6.1 is now available. I have installed > > it on Fedora 40 (via the normal dnf upgrade system) and look forward to > > testing it. > > Here's my experience after brief testing: > > * Apps are restored, including non-KDE ones such as Firefox and Evolution, > but Yakuake does not restore. > > * They all appear on the first desktop and have to be manually repositioned. > Window sizes are correct but window positioning within the desktop is > arbitrary. > > * AFAIK each app is responsible for its own internal state. Konsole does not > preserve tabs. Dolphin does not preserve terminal open/closed state > independently for each window (though all windows are present) and instead > of 3 windows in different directories they are all copies of each other > (probably a Dolphin problem). That's in line with comment 84. So still no proper session-restore for Plasma-Wayland. Clearly, this bug is not yet fixed and therefore should not be labelled as such. And to make Wayland the default for Plasma 6 with such a deficiency was a rather questionable decision which borders on self-sabotage, IMO. (In reply to imaginator from comment #125) > (In reply to Patrick O'Callaghan from comment #124) > > (In reply to Patrick O'Callaghan from comment #120) > > > Just a heads-up to note that Plasma 6.1 is now available. I have installed > > > it on Fedora 40 (via the normal dnf upgrade system) and look forward to > > > testing it. > > > > Here's my experience after brief testing: > > > > * Apps are restored, including non-KDE ones such as Firefox and Evolution, > > but Yakuake does not restore. > > > > * They all appear on the first desktop and have to be manually repositioned. > > Window sizes are correct but window positioning within the desktop is > > arbitrary. > > > > * AFAIK each app is responsible for its own internal state. Konsole does not > > preserve tabs. Dolphin does not preserve terminal open/closed state > > independently for each window (though all windows are present) and instead > > of 3 windows in different directories they are all copies of each other > > (probably a Dolphin problem). > > That's in line with comment 84. So still no proper session-restore for > Plasma-Wayland. Clearly, this bug is not yet fixed and therefore should not > be labelled as such. > > And to make Wayland the default for Plasma 6 with such a deficiency was a > rather questionable decision which borders on self-sabotage, IMO. Plasma under X11 didn't do those things correctly for me either, except possibly the tabs in Konsole. I'm not sure because I always used Autostart, which I'm intentionally not doing now, so behaviour maybe be different. (In reply to Patrick O'Callaghan from comment #126) > (In reply to imaginator from comment #125) > > (In reply to Patrick O'Callaghan from comment #124) > > > (In reply to Patrick O'Callaghan from comment #120) > > > > Just a heads-up to note that Plasma 6.1 is now available. I have installed > > > > it on Fedora 40 (via the normal dnf upgrade system) and look forward to > > > > testing it. > > > > > > Here's my experience after brief testing: > > > > > > * Apps are restored, including non-KDE ones such as Firefox and Evolution, > > > but Yakuake does not restore. > > > > > > * They all appear on the first desktop and have to be manually repositioned. > > > Window sizes are correct but window positioning within the desktop is > > > arbitrary. > > > > > > * AFAIK each app is responsible for its own internal state. Konsole does not > > > preserve tabs. Dolphin does not preserve terminal open/closed state > > > independently for each window (though all windows are present) and instead > > > of 3 windows in different directories they are all copies of each other > > > (probably a Dolphin problem). > > > > That's in line with comment 84. So still no proper session-restore for > > Plasma-Wayland. Clearly, this bug is not yet fixed and therefore should not > > be labelled as such. > > > > And to make Wayland the default for Plasma 6 with such a deficiency was a > > rather questionable decision which borders on self-sabotage, IMO. > > Plasma under X11 didn't do those things correctly for me either, except > possibly the tabs in Konsole. I'm not sure because I always used Autostart, > which I'm intentionally not doing now, so behaviour maybe be different. If session-restore also doesn't work in Plasma-6-X11 (it's always been working flawlessly for me in Plasma-5-X11) the situation would be even worse as there would be no correctly functioning alternative. For the sake of Plasma and KDE I hope that this is not the case. To give you an impression: among other things and apart from quite a few tabs in Dolphin and several instances of Konsole, Okular and Gwenview, I currently have over 20 docs open in Kate. That's not unusual for me. And would I want to re-open all that stuff _manually_ each time I have updated the kernel, for instance? In the year 2024!? I wouldn't even dream of it! (In reply to undoIT from comment #121) > Updated Arch with Plasma 6.1, rebooted, didn't work. Then, switched setting > to: "When session was manually saved" and rebooted. Don't remember seeing a > button to "Save Session" but it worked same as setting: "On last logout". > Switched back to "On last logout" and it is still working. Very pleased! > Thanks to everyone who put in the hard work to make this happen for Plasma 6. I am on Fedora 40 and I pretty much had to go through the same steps to actually enable this feature. Also the manual option didn't make the "save session" button show up even after reboot. System Info: Operating System: Fedora Linux 40 KDE Plasma Version: 6.1.0 KDE Frameworks Version: 6.3.0 Qt Version: 6.7.1 Kernel Version: 6.9.4-200.fc40.x86_64 (64-bit) Graphics Platform: Wayland Processors: 8 × Intel® Core™ i7-10510U CPU @ 1.80GHz Memory: 31.0 GiB of RAM Graphics Processor: Mesa Intel® UHD Graphics Manufacturer: LENOVO Product Name: 20RD002RUS System Version: ThinkPad E15 Fedora 40, KDE Plasma 6.1.0. When I login on a Wayland with my own account, I am greeted by a heap of 18 konsole and firefox windows, all stacked up on the first of my eight virtual desktops. I am the designated one to distribute and reposition them again, until next reboot. It's unusable like this. I can't use my own account under wayland, not even just for testing. The current state only really 'works' for one use case: you only have one desktop and all your windows are maximised. Granted, that's how most MS Windows users I know use their computer, so maybe it fills many people's needs. But it's not up to par with what a long time plasma/x11 user expects. Session management is very much _not_ fixed and because of this, plasma/wayland is still not ready for prime time in my workflow. Is there more work planned for plasma 6.2 to address the issues with session restore (via another bug)? or do we need to reopen this bug and address the remaining issues via this bug? (In reply to imaginator from comment #115) > So why don't you just dump Plasma-Wayland and use Plasma-X11 until this > essential (but nowadays rather trivial) feature is _fully_ implemented? Well, with modern laptops featuring extremely high resolution displays, and a multiscreen setup, X11 simply cannot handle such setups any more. With Wayland, a bunch of X11-features that have been "sort of working" like X-forwarding, screen sharing (in Teams/Zoom), and session management stopped working, but a whole lot of pain points that I had to work through over the past two decades went away, such as the resolution issue mentioned above. Ever since I adopted Wayland, screen sharing has been fixed and X-forwarding can be replaced by alternatives like VNC, which leaves session management as the last big ticket item. Regarding the fake session report, Fedora 40 has backported the patches to Plasma 6.0 a few months ago, so there was not really any change with the update to Plasma 6.1. I otherwise came to realise that the fake session restore is working exactly as advertised (cf. https://bugs.kde.org/show_bug.cgi?id=436318#c84 or the commit message at https://invent.kde.org/plasma/plasma-workspace/-/commit/660988b0e30ee8ccac98c0cf164b142d70709675). So, got to wait until somebody with the necessary knowledge and skills implements true session management in KDE Plasma. (In reply to Armin from comment #131) > (In reply to imaginator from comment #115) > > So why don't you just dump Plasma-Wayland and use Plasma-X11 until this > > essential (but nowadays rather trivial) feature is _fully_ implemented? > > Well, with modern laptops featuring extremely high resolution displays, and > a multiscreen setup, X11 simply cannot handle such setups any more. > > With Wayland, a bunch of X11-features that have been "sort of working" like > X-forwarding, screen sharing (in Teams/Zoom), and session management stopped > working, but a whole lot of pain points that I had to work through over the > past two decades went away, such as the resolution issue mentioned above. > Ever since I adopted Wayland, screen sharing has been fixed and X-forwarding > can be replaced by alternatives like VNC, which leaves session management as > the last big ticket item. > > Regarding the fake session report, Fedora 40 has backported the patches to > Plasma 6.0 a few months ago, so there was not really any change with the > update to Plasma 6.1. I otherwise came to realise that the fake session > restore is working exactly as advertised (cf. > https://bugs.kde.org/show_bug.cgi?id=436318#c84 or the commit message at > https://invent.kde.org/plasma/plasma-workspace/-/commit/ > 660988b0e30ee8ccac98c0cf164b142d70709675). > > So, got to wait until somebody with the necessary knowledge and skills > implements true session management in KDE Plasma. I generally agree, though I had hoped that the 6.1 version would be a little better and it really isn´t. What worries me is that apparently proper session restore is a missing feature in Wayland itself (Gnome has a non-standard workaround) so I don't know how long we are going to have to wait. This is not a good situation. I would be much happier if we could at least have windows restored to their proper desktops and positions, but I don't know enough to guess how hard that is to do. Is there no way for the 'session save' action to interrogate the window manager (or Wayland server) and save this information? I am also in the camp that windows not remembering which desktops were in is a dealbreaker for me. I sometimes use multiple desktops in my workflow and this feature may be more tedious to deal with than having session restore disabled. I'd rather have the feature disabled and start over from each reboot than have to sort through lots of windows to their right desktops. (In reply to Mustafa Kamran from comment #133) > I am also in the camp that windows not remembering which desktops were in is > a dealbreaker for me. I sometimes use multiple desktops in my workflow and > this feature may be more tedious to deal with than having session restore > disabled. I'd rather have the feature disabled and start over from each > reboot than have to sort through lots of windows to their right desktops. I defined some shortcuts (Ctrl-<N> for "send window to desktop N") but it's still tedious to have to do it for every login. (In reply to Patrick O'Callaghan from comment #132) > (In reply to Armin from comment #131) > > (In reply to imaginator from comment #115) > > > So why don't you just dump Plasma-Wayland and use Plasma-X11 until this > > > essential (but nowadays rather trivial) feature is _fully_ implemented? > > > > Well, with modern laptops featuring extremely high resolution displays, and > > a multiscreen setup, X11 simply cannot handle such setups any more. > > > > With Wayland, a bunch of X11-features that have been "sort of working" like > > X-forwarding, screen sharing (in Teams/Zoom), and session management stopped > > working, but a whole lot of pain points that I had to work through over the > > past two decades went away, such as the resolution issue mentioned above. > > Ever since I adopted Wayland, screen sharing has been fixed and X-forwarding > > can be replaced by alternatives like VNC, which leaves session management as > > the last big ticket item. > > > > Regarding the fake session report, Fedora 40 has backported the patches to > > Plasma 6.0 a few months ago, so there was not really any change with the > > update to Plasma 6.1. I otherwise came to realise that the fake session > > restore is working exactly as advertised (cf. > > https://bugs.kde.org/show_bug.cgi?id=436318#c84 or the commit message at > > https://invent.kde.org/plasma/plasma-workspace/-/commit/ > > 660988b0e30ee8ccac98c0cf164b142d70709675). > > > > So, got to wait until somebody with the necessary knowledge and skills > > implements true session management in KDE Plasma. > > I generally agree, though I had hoped that the 6.1 version would be a little > better and it really isn´t. What worries me is that apparently proper > session restore is a missing feature in Wayland itself (Gnome has a > non-standard workaround) so I don't know how long we are going to have to > wait. This is not a good situation. In this respect perhaps "Failand" might be a more suitable name. Anyway - in comment 25 I already pointed to the possibility of Wayland being the root-problem. And I guess that KDE has been aware of it since long (comment 62). But instead of using their clout as a leading Linux-DE and publicly pointing out and criticizing a glaring Wayland-deficiency they even chose Wayland as default for Plasma 6. IMO, this was a major strategic error as it not only leaves Plasma-6-users with a bad and obviously hard to fix regression but also lifted the pressure off the Wayland-devs to finally (!) provide a basic feature of modern DEs. Now it seems that Plasma-6-users are at the mercy of the Wayland-devs (comment 89). (In reply to imaginator from comment #135) > Now it seems that Plasma-6-users are at the mercy of the Wayland-devs > (comment 89). Well, not really. There is nothing stopping KDE devs to go ahead and implement the Wayland session management draft protocol. In fact, it would not be a bad thing at all to have an implementation of the protocol in order to identify any issues early on. The Gnome devs did exactly that, and implemented the draft protocol in Mutter and GTK: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3825 https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7370 Maybe this helps KDE devs to move things along for Plasma as well... (In reply to Armin from comment #136) > (In reply to imaginator from comment #135) > > Now it seems that Plasma-6-users are at the mercy of the Wayland-devs > > (comment 89). > > Well, not really. There is nothing stopping KDE devs to go ahead and > implement the Wayland session management draft protocol. In fact, it would > not be a bad thing at all to have an implementation of the protocol in order > to identify any issues early on. > > The Gnome devs did exactly that, and implemented the draft protocol in > Mutter and GTK: > https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3825 > https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7370 > > Maybe this helps KDE devs to move things along for Plasma as well... Taking the initiative and going ahead may indeed be the right move. Certainly better than wasting more time by waiting and hoping for a long overdue protocol to get approved (or not) at an unspecified point in time. (In reply to Nate Graham from comment #87) > Correct. Because real session restore is not implemented on Wayland yet, at > the moment this manual session restore hooks into the "real fake session > restore" mechanism that will be used until we get real session restore. It's a bit bad UX that this is not at least communicated to the user in the settings where this is selected. Quite frankly I cannot see why it was not hooked to the real session save which it at least attempts to do... Why was this bug marked resolved/fixed? It's still an issue under Wayland in plasma 6.1.1. Manually saving a session does not save all windows/apps and the ones that do reopen on the wrong desktops. Simply putting the onus on wayland to "hopefully" fix it is not a solution. Maybe we should change the severity of this bug to CRITICAL. Gentoo sort of forced me to update to latest plasma by deprecating some of the packages because of dependency hell. And now, the X11 session restore is broken....:-( (In reply to devsk from comment #141) > Gentoo sort of forced me to update to latest plasma by deprecating some of > the packages because of dependency hell. > > And now, the X11 session restore is broken....:-( Not sure about Gentoo, but for Fedora, there is no X11, Wayland is the default. X11 only apps are handled by a bridge. Even before this the Wayland session restore was broken. The X11 session restore was broken but was fixed some time ago. With Wayland some applications are not restored at all and onlers are all restored to the first virtual desktop. Had I known the session restore was not fixed in wayland I never would have upgraded to Fedora 40 from 39 (In reply to Brian Kaye from comment #142) The X11 package still exists for Fedora - see https://www.reddit.com/r/Fedora/comments/1cbgria/comment/l1dor96/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button - it's merely not included by default. (In reply to Brian Kaye from comment #142) > (In reply to devsk from comment #141) > > Gentoo sort of forced me to update to latest plasma by deprecating some of > > the packages because of dependency hell. > > > > And now, the X11 session restore is broken....:-( > > Not sure about Gentoo, but for Fedora, there is no X11, Wayland is the > default. X11 only apps are handled by a bridge. Even before this the Wayland > session restore was broken. The X11 session restore was broken but was fixed > some time ago. > > With Wayland some applications are not restored at all and onlers are all > restored to the first virtual desktop. Had I known the session restore was > not fixed in wayland I never would have upgraded to Fedora 40 from 39 That's Fedora's own initiative. Fedora's mission is to "lead the charge" on innovations even if it breaks some people's workflows. Plasma 6 still supports X11 session and there are more conservative distros that aim to support X11 (IDK if they upgraded to Plasma 6 yet though). (In reply to equeim from comment #144) openSUSE Tumbleweed has upgraded to Plasma 6, yet retained its XOrg 11 session. (In reply to Brian Kaye from comment #142) > (In reply to devsk from comment #141) > > Gentoo sort of forced me to update to latest plasma by deprecating some of > > the packages because of dependency hell. > > > > And now, the X11 session restore is broken....:-( > > Not sure about Gentoo, but for Fedora, there is no X11, Wayland is the > default. X11 only apps are handled by a bridge. Even before this the Wayland > session restore was broken. The X11 session restore was broken but was fixed > some time ago. > > With Wayland some applications are not restored at all and onlers are all > restored to the first virtual desktop. Had I known the session restore was > not fixed in wayland I never would have upgraded to Fedora 40 from 39 IIRC this was discussed on the Fedora KDE mailing list (by me and others) before F40 was released. I hate it but I go along with it because the alternative is to eventually fall out of support shortly after F41 comes out. I currently have shortcuts to move my (many) open windows to the right desktops. My main issue is with Dolphin not remembering the state of each of its open windows, but I suspect session restore on its own isn't going to fix that. See https://bugs.kde.org/show_bug.cgi?id=455324 Indeed, this is the last thing keeping me from using Wayland! I'd be happy with dolphin restoring for the beginning. (In reply to Mihai Sorin Dobrescu from comment #147) > Indeed, this is the last thing keeping me from using Wayland! > I'd be happy with dolphin restoring for the beginning. Dolphin doesn't restore properly on X11 either. I'm also having this problem in the X11-session of Plasma-5.27.11 (with security-update plasma-workspace-5.27.11.1). The symptoms are the same as described several times already. Just KDE-apps are restored but only in their native state, that is without previously opened documents etc. Now even the the latest version of Plasma 5 is completely unusable. What a dire story. (In reply to Patrick O'Callaghan from comment #148) > (In reply to Mihai Sorin Dobrescu from comment #147) > > Indeed, this is the last thing keeping me from using Wayland! > > I'd be happy with dolphin restoring for the beginning. > > Dolphin doesn't restore properly on X11 either. Currently, having the following configuration, Dolphin restores good enough, meaning it loses only its windows positions, but keeps all the opened paths, split views etc.. Operating System: MocaccinoOS (Gentoo) KDE Plasma Version: 5.27.11 KDE Frameworks Version: 5.116.0 Qt Version: 5.15.14 Kernel Version: 6.6.47-mocaccino (64-bit) Graphics Platform: X11 Processors: 16 × 11th Gen Intel® Core™ i7-11700K @ 3.60GHz Memory: 61.6 GiB of RAM Graphics Processor: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 Manufacturer: ASUS I've been following this bug because I used session restore back when it worked. I have found a workaround that works for me but I have not seen it mentioned here yet. Here's what I do. I have two monitors and use 4 virtual desktops. I'm able to have a console window with multiple tabs open on the left monitor in virtual desktop 1 by using the konsole option --tabs-from-file. One of the tabs starts with top running. Chrome starts on the left monitor in virtual desktop 2. Thunderbird starts on the right monitor in all virtual desktops. I am able to do this by using the autostart section of the system setting app to specify what apps start at login. Then in the Window Rules section (in Apps & Windows -> Window Management) I am able to specify where each app appears. Like I said, this works for me. Of course it does not save the current session so that it comes back on the next login. (In reply to Charles Dennett from comment #151) > I've been following this bug because I used session restore back when it > worked. I have found a workaround that works for me but I have not seen it > mentioned here yet. Here's what I do. I have two monitors and use 4 > virtual desktops. I'm able to have a console window with multiple tabs open > on the left monitor in virtual desktop 1 by using the konsole option > --tabs-from-file. One of the tabs starts with top running. Chrome starts > on the left monitor in virtual desktop 2. Thunderbird starts on the right > monitor in all virtual desktops. I am able to do this by using the > autostart section of the system setting app to specify what apps start at > login. Then in the Window Rules section (in Apps & Windows -> Window > Management) I am able to specify where each app appears. Like I said, this > works for me. Of course it does not save the current session so that it > comes back on the next login. AFAIK this wouldn't solve the "multiple Dolphin windows" problem. To summarise: I have 3 Dolphin windows, all on the same desktop. Window A has two panes, each in a different directory. Window B also has two panes, each in a different directory. Window C has a single pane but an open terminal and is in yet another directory. On session restore, all three windows are identical copies of *one* of A, B or C. Which one it happens to be is essentially random, though it might be the one that happened to be on top last time. When I first reported this over two years ago I was using X11. AFAIK the bug has received no attention in the interim. (This should really be discussed on the other report already linked above). (In reply to Patrick O'Callaghan from comment #152) > (In reply to Charles Dennett from comment #151) > > I've been following this bug because I used session restore back when it > > worked. I have found a workaround that works for me but I have not seen it > > mentioned here yet. Here's what I do. I have two monitors and use 4 > > virtual desktops. I'm able to have a console window with multiple tabs open > > on the left monitor in virtual desktop 1 by using the konsole option > > --tabs-from-file. One of the tabs starts with top running. Chrome starts > > on the left monitor in virtual desktop 2. Thunderbird starts on the right > > monitor in all virtual desktops. I am able to do this by using the > > autostart section of the system setting app to specify what apps start at > > login. Then in the Window Rules section (in Apps & Windows -> Window > > Management) I am able to specify where each app appears. Like I said, this > > works for me. Of course it does not save the current session so that it > > comes back on the next login. > > AFAIK this wouldn't solve the "multiple Dolphin windows" problem. To > summarise: > > I have 3 Dolphin windows, all on the same desktop. Window A has two panes, > each in a different directory. Window B also has two panes, each in a > different directory. Window C has a single pane but an open terminal and is > in yet another directory. > > On session restore, all three windows are identical copies of *one* of A, B > or C. Which one it happens to be is essentially random, though it might be > the one that happened to be on top last time. "Thunar" keeps the settings after (Plasma-) session-restore, just checked your use case. It doesn't have an integrated terminal, though. (In reply to imaginator from comment #153) [...] > > I have 3 Dolphin windows, all on the same desktop. Window A has two panes, > > each in a different directory. Window B also has two panes, each in a > > different directory. Window C has a single pane but an open terminal and is > > in yet another directory. > > > > On session restore, all three windows are identical copies of *one* of A, B > > or C. Which one it happens to be is essentially random, though it might be > > the one that happened to be on top last time. > "Thunar" keeps the settings after (Plasma-) session-restore, just checked > your use case. It doesn't have an integrated terminal, though. I just tried it, but no. After saving the session and re-logging in, Thunar is not restored and when I start it manually it doesn't recover the saved sessions. I haven't tried it with Auto-start. I note that it does have an "open terminal here" option (which currently fails because of some setting). I presume it doesn't open a new pane in the same window, though that wouldn't be critical. (In reply to Patrick O'Callaghan from comment #154) > (In reply to imaginator from comment #153) > [...] > > > I have 3 Dolphin windows, all on the same desktop. Window A has two panes, > > > each in a different directory. Window B also has two panes, each in a > > > different directory. Window C has a single pane but an open terminal and is > > > in yet another directory. > > > > > > On session restore, all three windows are identical copies of *one* of A, B > > > or C. Which one it happens to be is essentially random, though it might be > > > the one that happened to be on top last time. > > "Thunar" keeps the settings after (Plasma-) session-restore, just checked > > your use case. It doesn't have an integrated terminal, though. > > I just tried it, but no. After saving the session and re-logging in, Thunar > is not restored and when I start it manually it doesn't recover the saved > sessions. I haven't tried it with Auto-start. I note that it does have an Sorry. Perhaps we are using different Plasma-versions. I reverted to -5.27.11 where session-restore does work for X11 (it does _not_ in -5.27.11.1). And here Thunar-4.18.6 behaves as described. > "open terminal here" option (which currently fails because of some setting). > I presume it doesn't open a new pane in the same window, though that > wouldn't be critical. Actually Thunar does offer split windows. And for the one window where you need the integrated terminal you could use Dolphin. Good luck! (In reply to imaginator from comment #155) [...] > > "open terminal here" option (which currently fails because of some setting). > > I presume it doesn't open a new pane in the same window, though that > > wouldn't be critical. > Actually Thunar does offer split windows. And for the one window where you > need the integrated terminal you could use Dolphin. Yes, I know Thunar has horizontally split panes. I meant that a terminal wouldn't open in a new horizontal pane in the same window, as with Dolphin. However this is a minor point. Git commit 4dff1973116597210cf9fb1c102e5c9433d13b1d by Fushan Wen, on behalf of David Edmundson. Committed on 14/09/2024 at 13:22. Pushed by fusionfuture into branch 'master'. startkde: Fix wayland session restore saving When we added manual saving support to plasma-shutdown a guard was changed in the normal shutdown path. This guard was wrong, we want to check we're in the restorePreviousLogout path here. This amends 804976c5ecec1fbf5f6e7e09970a8269bdf748d2 M +1 -1 startkde/plasma-shutdown/shutdown.cpp https://invent.kde.org/plasma/plasma-workspace/-/commit/4dff1973116597210cf9fb1c102e5c9433d13b1d (In reply to Fushan Wen from comment #157) > Git commit 4dff1973116597210cf9fb1c102e5c9433d13b1d by Fushan Wen, on behalf > of David Edmundson. > Committed on 14/09/2024 at 13:22. > Pushed by fusionfuture into branch 'master'. > > startkde: Fix wayland session restore saving > > When we added manual saving support to plasma-shutdown a guard was > changed in the normal shutdown path. This guard was wrong, we want to > check we're in the restorePreviousLogout path here. > > This amends 804976c5ecec1fbf5f6e7e09970a8269bdf748d2 > > M +1 -1 startkde/plasma-shutdown/shutdown.cpp > > https://invent.kde.org/plasma/plasma-workspace/-/commit/ > 4dff1973116597210cf9fb1c102e5c9433d13b1d Hi, I've just tried this on Plasma 6.2 Dev on KDE Neon. The build was done from scratch issuing kde-builder --refresh-build workspace --ignore-modules sddm-kcm (SDDM KCM module fails at build but it's not essential here). I've got the following behavior: - left a Dolphin open at logoff/shutdown - logged in back with a Wayland session - Dolphin opened at login, but maximized on the screen - left a Dolphin open at logoff/shutdown once more - logged in back - no Dolphin anymore Git commit fae71841ff018e34bbb02aa1098b53f63c1743d2 by Fushan Wen. Committed on 17/09/2024 at 07:53. Pushed by fusionfuture into branch 'Plasma/6.2'. startkde: Fix wayland session restore saving When we added manual saving support to plasma-shutdown a guard was changed in the normal shutdown path. This guard was wrong, we want to check we're in the restorePreviousLogout path here. This amends 804976c5ecec1fbf5f6e7e09970a8269bdf748d2 (cherry picked from commit 4dff1973116597210cf9fb1c102e5c9433d13b1d) 89ee67ad startkde: Fix wayland session restore saving Co-authored-by: David Edmundson <kde@davidedmundson.co.uk> M +1 -1 startkde/plasma-shutdown/shutdown.cpp https://invent.kde.org/plasma/plasma-workspace/-/commit/fae71841ff018e34bbb02aa1098b53f63c1743d2 so is it fixed now? If I install kubuntu 24.04, and the necessary updates, will I have full session restore ? On 17/09/24 17:46, Nate Graham wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > Nate Graham <nate@kde.org> changed: > > What |Removed |Added > ---------------------------------------------------------------------------- > Version Fixed In|6.2 |6.2.0 > If Kubuntu 24.04 ships Plasma 6.2.0, then yes, unless the change that fixed this didn't actually work. what distro should I install to get session restore working? or can I install Plasma 6.2.0 on top of Kubuntu 24.04 ? On 18/09/24 14:06, Nate Graham wrote: > https://bugs.kde.org/show_bug.cgi?id=436318 > > --- Comment #161 from Nate Graham <nate@kde.org> --- > If Kubuntu 24.04 ships Plasma 6.2.0, then yes, unless the change that fixed > this didn't actually work. > There are over 100 people CCd on this bug report, and every comment sends all of them an email. So I would recommend taking installation and versioning questions like those elsewhere — for example to https://discuss.kde.org. Thanks a lot! I wonder why there is extra button for save session since this is the button for leave session when configuration is restore previous session? Am I correct in assuming that the commit marked as fixing this issue only enables the 'fake session restore' (that opens all the applications that were previously opened without restoring their state)? From my experiments, it still looks like *saving* the session information (which is what this bug title references) is not fixed by that commit. That would probably require progress on https://invent.kde.org/plasma/kwin/-/issues/113 (In reply to Kishore Gopalakrishnan from comment #165) > Am I correct in assuming that the commit marked as fixing this issue only > enables the 'fake session restore' (that opens all the applications that > were previously opened without restoring their state)? > > From my experiments, it still looks like *saving* the session information > (which is what this bug title references) is not fixed by that commit. That > would probably require progress on > https://invent.kde.org/plasma/kwin/-/issues/113 From [1]: "Fixed a bug that caused real-fake-session-restore to not work properly on Wayland (David Edmundson, Plasma 6.2.0. Link)" I'm not sure what a "real-fake-session-restore" is but I guess it's the surrogate session-restore that's been delivered to Plasma-6-wayland-users lately. Now, if only that crutch is fixed, it's indeed hard to see how that would qualify to mark *this* bug (again) as "resolved". Perhaps after three years the pain became just too much to bear. -- [1] https://pointieststick.com/2024/09/20/this-week-in-plasma-polishing-like-mad/ (In reply to imaginator from comment #166) > (In reply to Kishore Gopalakrishnan from comment #165) > > Am I correct in assuming that the commit marked as fixing this issue only > > enables the 'fake session restore' (that opens all the applications that > > were previously opened without restoring their state)? > > > > From my experiments, it still looks like *saving* the session information > > (which is what this bug title references) is not fixed by that commit. That > > would probably require progress on > > https://invent.kde.org/plasma/kwin/-/issues/113 > > From [1]: "Fixed a bug that caused real-fake-session-restore to not work > properly on Wayland (David Edmundson, Plasma 6.2.0. Link)" > > I'm not sure what a "real-fake-session-restore" is but I guess it's the > surrogate session-restore that's been delivered to Plasma-6-wayland-users > lately. > > Now, if only that crutch is fixed, it's indeed hard to see how that would > qualify to mark *this* bug (again) as "resolved". Perhaps after three years > the pain became just too much to bear. IMHO the bug is definitely not Resolved. Save/Restore session continues to be an unimplemented feature. I appreciate Nate's efforts with regard to "fake session restore" but it is a stopgap, not a genuine solution. (In reply to Patrick O'Callaghan from comment #167) > (In reply to imaginator from comment #166) > > (In reply to Kishore Gopalakrishnan from comment #165) > > > Am I correct in assuming that the commit marked as fixing this issue only > > > enables the 'fake session restore' (that opens all the applications that > > > were previously opened without restoring their state)? > > > > > > From my experiments, it still looks like *saving* the session information > > > (which is what this bug title references) is not fixed by that commit. That > > > would probably require progress on > > > https://invent.kde.org/plasma/kwin/-/issues/113 > > > > From [1]: "Fixed a bug that caused real-fake-session-restore to not work > > properly on Wayland (David Edmundson, Plasma 6.2.0. Link)" > > > > I'm not sure what a "real-fake-session-restore" is but I guess it's the > > surrogate session-restore that's been delivered to Plasma-6-wayland-users > > lately. > > > > Now, if only that crutch is fixed, it's indeed hard to see how that would > > qualify to mark *this* bug (again) as "resolved". Perhaps after three years > > the pain became just too much to bear. > > IMHO the bug is definitely not Resolved. Save/Restore session continues to > be an unimplemented feature. I appreciate Nate's efforts with regard to > "fake session restore" but it is a stopgap, not a genuine solution. Well, you reported this bug. And if you think that it is still not resolved then re-open it. I'm pretty sure that the vast majority of people on the cc-list would support and appreciate that move. However, it would be better if the assignee(s) reopened it. Doing so would be the most sensible thing anyway for anyone who has understood what this bug-report is about. It would also display some respect for their users and that they are not trying to lead them down the garden path. (In reply to imaginator from comment #168) > (In reply to Patrick O'Callaghan from comment #167) > > (In reply to imaginator from comment #166) > > > (In reply to Kishore Gopalakrishnan from comment #165) > > > > Am I correct in assuming that the commit marked as fixing this issue only > > > > enables the 'fake session restore' (that opens all the applications that > > > > were previously opened without restoring their state)? > > > > > > > > From my experiments, it still looks like *saving* the session information > > > > (which is what this bug title references) is not fixed by that commit. That > > > > would probably require progress on > > > > https://invent.kde.org/plasma/kwin/-/issues/113 > > > > > > From [1]: "Fixed a bug that caused real-fake-session-restore to not work > > > properly on Wayland (David Edmundson, Plasma 6.2.0. Link)" > > > > > > I'm not sure what a "real-fake-session-restore" is but I guess it's the > > > surrogate session-restore that's been delivered to Plasma-6-wayland-users > > > lately. > > > > > > Now, if only that crutch is fixed, it's indeed hard to see how that would > > > qualify to mark *this* bug (again) as "resolved". Perhaps after three years > > > the pain became just too much to bear. > > > > IMHO the bug is definitely not Resolved. Save/Restore session continues to > > be an unimplemented feature. I appreciate Nate's efforts with regard to > > "fake session restore" but it is a stopgap, not a genuine solution. > > Well, you reported this bug. And if you think that it is still not resolved > then re-open it. I'm pretty sure that the vast majority of people on the > cc-list would support and appreciate that move. > > However, it would be better if the assignee(s) reopened it. Doing so would > be the most sensible thing anyway for anyone who has understood what this > bug-report is about. It would also display some respect for their users and > that they are not trying to lead them down the garden path. Done. I also edited the title to make it clear that it covers Resume as well as Save. Session restore still not fixed with plasma 6.2 rpms on Fedora FC40. Still not fixed with plasma 6.2.3-1 rpms on Fedora 31. Sessions still restored to first virtual desktop. A pain but they are at least restored somewhere. That's something else: Bug 482816. (In reply to Brian Kaye from comment #171) > Still not fixed with plasma 6.2.3-1 rpms on Fedora 31. Sessions still > restored to first virtual desktop. A pain but they are at least restored > somewhere. Typo. Shoul refer to Fedora 31 rpms Should be Fedora 41 RPMS. I am a lousy typer :-). Would be nice to edit a comment. Folks, please try to keep down the chatter in here. I know it's very frustrating that this feature hasn't been implemented on Wayland yet (it is for me too, believe me), but more comments in here about how Wayland sucks or KDE developers are like Microsoft devs or workarounds using other software are off-topic, and I have hidden them to improve the signal-to-noise ratio in here. Let's please try to keep comments technical and steered in the direction of getting the feature implemented. Thanks! (In reply to Nate Graham from comment #175) Thanks, Nate, for confirming that KDE developers are working on this. I guess users would be less confused if you also clean the "Version fixed in" filed, because it is clearly not fixed and most of the recent comments say just that. Good call, I forgot to do that. Thanks. *** Bug 496674 has been marked as a duplicate of this bug. *** *** Bug 471977 has been marked as a duplicate of this bug. *** I'd like to add to this bug report. I can reliably (100% of the time) reproduce this bug as submitted on Fedora 41 KDE Frameworks Version: 6.9.0 Qt Version: 6.8.1 Graphics Platform: Wayland Additionally, once I select "Application Launcher" -> Leave -> "Save Session" CTRL-ALT-DEL becomes 'inactive.' That is, hitting CTRL-ALT-DEL has no response by the system whatsoever. To restart or shutdown I have to use the bash command line. I can also reproduce this behavior 100% of the time. I currently have a set of apps in Autostart, but it's still annoying to have to manually move windows onto the right desktops after logging in. As a partial workaround, I wrote a couple of scripts using kdotool (see your local KDE repo). They can occasionally give strange error messages, and don't do anything for window positions within a desktop because that doesn't seem to be possible, but I thought I'd post them here in case anyone finds them useful. Note that I invoke them manually. There may be a way to automate this, but I haven't found it. (Simple login/logout scripts are not the answer).: Save: -------- #!/bin/bash # Save desktop state of selected windows # pocallaghan@gmail.com - 22/11/2024 # FIXME: assumes all the window names are distinct # and restored on restarting SAVE_FILE=$HOME/.save_desktops true > "$SAVE_FILE" # Zero the file apps=(dolphin firefox evolution konsole) for app in "${apps[@]}" do for d in $(kdotool search "$app") do # cn=$(kdotool getwindowclassname $d) # [ "$cn" = plasmashell ] && continue # Plasma has no desktop dt=$(kdotool get_desktop_for_window "$d") wn=$(kdotool getwindowname "$d") echo "$dt $wn" >> "$SAVE_FILE" done done --------- Restore: --------- #!/bin/bash # Restore the virtual desktops of saved windows # pocallaghan@gmail.com - 23/11/2024 SAVE_FILE=$HOME/.save_desktops cat "$SAVE_FILE"|while read dt wn ; do window=$(kdotool search "$wn") [ "$window"x = x ] && continue kdotool set_desktop_for_window "$window" "$dt" done ------ (In reply to Patrick O'Callaghan from comment #181) If you uploaded the script to even something as simple as a GitHub Gist or GitLab Snippet, and then linked that here, I think it'd be more useful, since questions and suggestions can't be posted here without notifying 118 people. Thanks for it, though. (In reply to Roke Julian Lockhart Beedell from comment #182) > (In reply to Patrick O'Callaghan from comment #181) > If you uploaded the script to even something as simple as a GitHub Gist or > GitLab Snippet, and then linked that here, I think it'd be more useful, > since questions and suggestions can't be posted here without notifying 118 > people. Thanks for it, though. Good idea. I've put it on: https://gist.github.com/pjoc/a5059b93321ee67d42f4af8d45b7276f Thanks While hunting for any recent updates, I found this. Looks like gnome-48 will contain some form of session restore: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3825 https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7682 https://www.phoronix.com/news/Mutter-XDG-Session-Management I can't really find any recent updates for KDE for similar effort. Is this ever going to be worked on by KDE? (In reply to devsk from comment #184) > While hunting for any recent updates, I found this. Looks like gnome-48 will > contain some form of session restore: > > https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3825 > https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7682 > https://www.phoronix.com/news/Mutter-XDG-Session-Management > > I can't really find any recent updates for KDE for similar effort. Is this > ever going to be worked on by KDE? I surely hope so. I'm sure I'm not the only one who finds it extremely annoying to have to reposition my windows each time I log in or reboot my system. In fact, I update my system only once a month (dnf upgrade) because I'm tired of having to resize and reposition my windows each time. SunOS OpenLook did this beautifully, accurately, and reliably way back in 1990. I've tried 4 Linux distros, and this does not work on any of the ones I've tried. Why is it that Linux has so much trouble with basic things like this? Who does the prioritization of work on Linux platforms? Do they get the feedback from users? > Is this ever going to be worked on by KDE? In my mind, there are two answers to this question, and both are correct answers: 1. We have no plan to get this working on KDE anytime soon 2. We have a plan and here is the schedule (gnome guys took this approach) I haven't seen 2) for KDE anywhere on the interwebs. Can we at least create a plan? Its fine if its done in 6 months from now or 1 year. But let's have a plan and work toward it. 1) is also fine but then, let's state that clearly so that we can stop coming back to this bug (its almost 4 years old now) every few months. To me, it almost feels like, because this bug is open, someone is working on it in the background when, in reality, no one is, and there is no way of knowing if someone is or not. PS: I found the following for kwin but its 2 years old and open: https://invent.kde.org/plasma/kwin/-/merge_requests/3024 (In reply to Peter Huatan from comment #185) > > Is this ever going to be worked on by KDE? > > I surely hope so. I'm sure I'm not the only one who finds it extremely > annoying to have to reposition my windows each time I log in or reboot my > system. In fact, I update my system only once a month (dnf upgrade) because > I'm tired of having to resize and reposition my windows each time. There is a work-round if you use sddm: select X11 on the login screen, rather than Wayland, and you'll have the proper restoration of desktops and their windows. It works for me, anyway. (In reply to Peter Humphrey from comment #187) > (In reply to Peter Huatan from comment #185) > > > > Is this ever going to be worked on by KDE? > > > > I surely hope so. I'm sure I'm not the only one who finds it extremely > > annoying to have to reposition my windows each time I log in or reboot my > > system. In fact, I update my system only once a month (dnf upgrade) because > > I'm tired of having to resize and reposition my windows each time. > > There is a work-round if you use sddm: select X11 on the login screen, > rather than Wayland, and you'll have the proper restoration of desktops and > their windows. It works for me, anyway. Regarding session-restore, Plasma-X11 works for anybody. ;) The question is whether your distribution offers it, as Plasma-wayland unfortunately is the default now. To me the central problem in this affair seems to be the obviously still missing wayland-protocol (see Comment 89). (In reply to imaginator from comment #188) > (In reply to Peter Humphrey from comment #187) > > (In reply to Peter Huatan from comment #185) > > > > > > Is this ever going to be worked on by KDE? > > > > > > I surely hope so. I'm sure I'm not the only one who finds it extremely > > > annoying to have to reposition my windows each time I log in or reboot my > > > system. In fact, I update my system only once a month (dnf upgrade) because > > > I'm tired of having to resize and reposition my windows each time. > > > > There is a work-round if you use sddm: select X11 on the login screen, > > rather than Wayland, and you'll have the proper restoration of desktops and > > their windows. It works for me, anyway. > > Regarding session-restore, Plasma-X11 works for anybody. ;) > The question is whether your distribution offers it, as Plasma-wayland > unfortunately is the default now. > > To me the central problem in this affair seems to be the obviously still > missing wayland-protocol (see Comment 89). I said: "if you use sddm." (In reply to Peter Humphrey from comment #189) > (In reply to imaginator from comment #188) > > (In reply to Peter Humphrey from comment #187) > > > (In reply to Peter Huatan from comment #185) > > > > > > > > Is this ever going to be worked on by KDE? > > > > > > > > I surely hope so. I'm sure I'm not the only one who finds it extremely > > > > annoying to have to reposition my windows each time I log in or reboot my > > > > system. In fact, I update my system only once a month (dnf upgrade) because > > > > I'm tired of having to resize and reposition my windows each time. > > > > > > There is a work-round if you use sddm: select X11 on the login screen, > > > rather than Wayland, and you'll have the proper restoration of desktops and > > > their windows. It works for me, anyway. > > > > Regarding session-restore, Plasma-X11 works for anybody. ;) > > The question is whether your distribution offers it, as Plasma-wayland > > unfortunately is the default now. > > > > To me the central problem in this affair seems to be the obviously still > > missing wayland-protocol (see Comment 89). > > I said: "if you use sddm." It doesn't matter whether your display manager is sddm, lightdm, gdm or whatever. If the distribution supports Plasma-X11 in addition to Plasma-Wayland, you can choose between them when logging in. (In reply to Peter Huatan from comment #185) > (In reply to devsk from comment #184) > > While hunting for any recent updates, I found this. Looks like gnome-48 will > > contain some form of session restore: > > > > https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3825 > > https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7682 > > https://www.phoronix.com/news/Mutter-XDG-Session-Management > > > > I can't really find any recent updates for KDE for similar effort. Is this > > ever going to be worked on by KDE? > > I surely hope so. I'm sure I'm not the only one who finds it extremely > annoying to have to reposition my windows each time I log in or reboot my > system. In fact, I update my system only once a month (dnf upgrade) because > I'm tired of having to resize and reposition my windows each time. > > SunOS OpenLook did this beautifully, accurately, and reliably way back in > 1990. I've tried 4 Linux distros, and this does not work on any of the ones > I've tried. Why is it that Linux has so much trouble with basic things like > this? Who does the prioritization of work on Linux platforms? Do they get > the feedback from users? A Linux-system is a huge mosaic of several hundreds or more pieces of software provided by probably as many individual projects (communities of developers). There is no central authority that can set/enforce priorities or order certain devs to get out of slo-mo mode and finally provide what others urgently need. By making Wayland the default for Plasma-6 instead of sticking with X11 (and publicly explaining why), KDE unfortunately gave away the only leverage they had in this affair. (In reply to imaginator from comment #191) > By making Wayland the default for Plasma-6 instead of sticking with X11 (and > publicly explaining why), KDE unfortunately gave away the only leverage they > had in this affair. openSUSE Leap 16.x still uses X11 / KDE 5 as default as far as I know or better as I'm using it. (In reply to Andreas Hartmann from comment #192) > (In reply to imaginator from comment #191) > > By making Wayland the default for Plasma-6 instead of sticking with X11 (and > > publicly explaining why), KDE unfortunately gave away the only leverage they > > had in this affair. > > openSUSE Leap 16.x still uses X11 / KDE 5 as default as far as I know or > better as I'm using it. You probably mean 15.x? AFAIK they do not even offer Wayland for those versions. But anyway, we're talking about Plasma-6 here, not Plasma-5. KDE chose Wayland as default for Plasma-6. So if a distribution only offers the default Plasma-6, then the user has no choice - and no session-restore. (In reply to Patrick O'Callaghan from comment #183) > (In reply to Roke Julian Lockhart Beedell from comment #182) > > (In reply to Patrick O'Callaghan from comment #181) > > If you uploaded the script to even something as simple as a GitHub Gist or > > GitLab Snippet, and then linked that here, I think it'd be more useful, > > since questions and suggestions can't be posted here without notifying 118 > > people. Thanks for it, though. > > Good idea. I've put it on: > > https://gist.github.com/pjoc/a5059b93321ee67d42f4af8d45b7276f > > Thanks Updated the Restore script to handle window titles with special characters. (In reply to Patrick O'Callaghan from comment #181) > I currently have a set of apps in Autostart, but it's still annoying to have > to manually move windows onto the right desktops after logging in. As a > partial workaround, I wrote a couple of scripts using kdotool (see your > local KDE repo). They can occasionally give strange error messages, and > don't do anything for window positions within a desktop because that doesn't > seem to be possible, but I thought I'd post them here in case anyone finds > them useful. Note that I invoke them manually. There may be a way to > automate this, but I haven't found it. (Simple login/logout scripts are not > the answer).: Have you ever put the script that restores the windows into "Autostart" as well? I don't know whether one has any influence on the sequence in which the programs are started but _if_ they are started in alphabetical order you could name it something like z_script. More on topic: the funny and real mean thing with session-save/-restore in Plasma-wayland (-5.27.11.1) is that _restoring_ a previous session does work fine (perhaps apart from minor differences in window-positions). I can tell because when I log out of a Plasma-X11 session and then log in to a Plasma-wayland session, the previous (Plasma-X11) session is restored almost perfectly, incl. kate, dolphin, whatever. And that's reproducible. However, if one then logs out of the Plasma-wayland session and in to a Plasma-X11 session (or Plasma-wayland session), the previous session is gone, except for non-KDE-apps like Firefox or Thunderbird. So, from what I see here with Plasma-wayland, the real problem is not _restoring_ a session but _saving_ it correctly. And what makes matters worse (somehow) is that the state of non-KDE-apps like Firefox/Thunderbird _is_ saved correctly in ksmserverrc [1]. What's not saved is the state of KDE-apps. But perhaps you know all this already. ;) -- [1] [LegacySession: saved at previous logout] count=0 [LegacySession: saved by user] count=0 [Session: saved at previous logout] clientId1=104c55580000173762586300000023360009 clientId2=104c55580000173870527300000023170111 clientId3=104c55580000173902919800000078330002 count=3 program1=thunderbird program2=firefox program3=pulseaudio restartCommand1=/opt/thunderbird/thunderbird,--sm-client-id,104c55580000173762586300000023360009 restartCommand2=/opt/firefox/firefox,--sm-client-id,104c55580000173870527300000023170111 restartCommand3= [...] (In reply to imaginator from comment #195) > (In reply to Patrick O'Callaghan from comment #181) > > I currently have a set of apps in Autostart, but it's still annoying to have > > to manually move windows onto the right desktops after logging in. As a > > partial workaround, I wrote a couple of scripts using kdotool (see your > > local KDE repo). They can occasionally give strange error messages, and > > don't do anything for window positions within a desktop because that doesn't > > seem to be possible, but I thought I'd post them here in case anyone finds > > them useful. Note that I invoke them manually. There may be a way to > > automate this, but I haven't found it. (Simple login/logout scripts are not > > the answer).: > > Have you ever put the script that restores the windows into "Autostart" as > well? I don't know whether one has any influence on the sequence in which > the programs are started but _if_ they are started in alphabetical order you > could name it something like z_script. It doesn't work (I tried it). AFAIK Plasma fires all the Autostart scripts concurrently and I don't see a way to make the Restore script wait until everything else is up, or at least has mapped its window(s). Maybe there's some systemd magic that would do it. > More on topic: the funny and real mean thing with session-save/-restore in > Plasma-wayland (-5.27.11.1) is that _restoring_ a previous session does work > fine (perhaps apart from minor differences in window-positions). I can tell > because when I log out of a Plasma-X11 session and then log in to a > Plasma-wayland session, the previous (Plasma-X11) session is restored almost > perfectly, incl. kate, dolphin, whatever. And that's reproducible. > However, if one then logs out of the Plasma-wayland session and in to a > Plasma-X11 session (or Plasma-wayland session), the previous session is > gone, except for non-KDE-apps like Firefox or Thunderbird. > > So, from what I see here with Plasma-wayland, the real problem is not > _restoring_ a session but _saving_ it correctly. And what makes matters > worse (somehow) is that the state of non-KDE-apps like Firefox/Thunderbird > _is_ saved correctly in ksmserverrc [1]. What's not saved is the state of > KDE-apps. But perhaps you know all this already. ;) That is interesting. (In reply to Patrick O'Callaghan from comment #196) > (In reply to imaginator from comment #195) > > (In reply to Patrick O'Callaghan from comment #181) > > > I currently have a set of apps in Autostart, but it's still annoying to have > > > to manually move windows onto the right desktops after logging in. As a > > > partial workaround, I wrote a couple of scripts using kdotool (see your > > > local KDE repo). They can occasionally give strange error messages, and > > > don't do anything for window positions within a desktop because that doesn't > > > seem to be possible, but I thought I'd post them here in case anyone finds > > > them useful. Note that I invoke them manually. There may be a way to > > > automate this, but I haven't found it. (Simple login/logout scripts are not > > > the answer).: > > > > Have you ever put the script that restores the windows into "Autostart" as > > well? I don't know whether one has any influence on the sequence in which > > the programs are started but _if_ they are started in alphabetical order you > > could name it something like z_script. > > It doesn't work (I tried it). AFAIK Plasma fires all the Autostart scripts > concurrently and I don't see a way to make the Restore script wait until > everything else is up, or at least has mapped its window(s). Maybe there's > some systemd magic that would do it. You could use "sleep" (-> sleep --help) to delay the execution of the relevant parts of your script until other scripts have run or the desktop is fired up. Example: #!/bin/bash # Test "sleep". sleep 10 echo "This was a test." HTH. (In reply to imaginator from comment #197) On older systems, this wouldn't really work, since there's always something that randomly takes ages to launch. Is there a way to wait for a specific process to be invoked? If so, perhaps wait for `plasmashell` to be running, then run the script, since it's usually one of the last things to run. https://stackoverflow.com/a/9118509/9731176 might be of use. (In reply to imaginator from comment #197) > (In reply to Patrick O'Callaghan from comment #196) > > (In reply to imaginator from comment #195) > > > (In reply to Patrick O'Callaghan from comment #181) > > > > I currently have a set of apps in Autostart, but it's still annoying to have > > > > to manually move windows onto the right desktops after logging in. As a > > > > partial workaround, I wrote a couple of scripts using kdotool (see your > > > > local KDE repo). They can occasionally give strange error messages, and > > > > don't do anything for window positions within a desktop because that doesn't > > > > seem to be possible, but I thought I'd post them here in case anyone finds > > > > them useful. Note that I invoke them manually. There may be a way to > > > > automate this, but I haven't found it. (Simple login/logout scripts are not > > > > the answer).: > > > > > > Have you ever put the script that restores the windows into "Autostart" as > > > well? I don't know whether one has any influence on the sequence in which > > > the programs are started but _if_ they are started in alphabetical order you > > > could name it something like z_script. > > > > It doesn't work (I tried it). AFAIK Plasma fires all the Autostart scripts > > concurrently and I don't see a way to make the Restore script wait until > > everything else is up, or at least has mapped its window(s). Maybe there's > > some systemd magic that would do it. > > You could use "sleep" (-> sleep --help) to delay the execution of the > relevant parts of your script until other scripts have run or the desktop is > fired up. Example: > > #!/bin/bash > > # Test "sleep". > sleep 10 > echo "This was a test." > > > HTH. I generally think timer-based solutions are inferior to triggers, but you can do that if want. It strikes me as even more kludgey than what I already have, and given that the user has to manually save the session anyway it wouldn't even have the advantage of transparency. Behaviour may have changed. Not sure of when this happened but I just noticed that the session restore does not resume windows to the first virtual desktop. It restores the windows to the last virtual desktop that was used in the previous session. I little window is flashed that says the name of the virtual desktop and then proceeds to restore the windows in that desktop. I finally moved to wayland with 6.3. I had to jump through some hoops though. Here is the code in case someone wants to try the save/restore scripts (make sure to read the comments on the gist before using this code): https://gist.github.com/devsk/9eff2d08d6cd5772afd2b178930b66ae (In reply to Roke Julian Lockhart Beedell from comment #198) > (In reply to imaginator from comment #197) > On older systems, this wouldn't really work, since there's always something > that randomly takes ages to launch. Is there a way to wait for a specific > process to be invoked? If so, perhaps wait for `plasmashell` to be running, > then run the script, since it's usually one of the last things to run. > https://stackoverflow.com/a/9118509/9731176 might be of use. I simply went with the option of starting all the applications from the saved session and then giving user a dialog box to OK the moving of the windows to their respective desktops, positioning them at the saved location and resizing them as per the saved info. The user just waits for all the application windows to start in the current desktop and then, pressing OK does the work. There are tricks to center the dialog box, keep it above all windows and activating it. kdotool works really well to do the rest. Checkout the gist I posted. Most of code can be used as is. Only binary names need adjusting/addition/deletion. (In reply to devsk from comment #202) > (In reply to Roke Julian Lockhart Beedell from comment #198) > > (In reply to imaginator from comment #197) > > On older systems, this wouldn't really work, since there's always something > > that randomly takes ages to launch. Is there a way to wait for a specific > > process to be invoked? If so, perhaps wait for `plasmashell` to be running, > > then run the script, since it's usually one of the last things to run. > > https://stackoverflow.com/a/9118509/9731176 might be of use. > > I simply went with the option of starting all the applications from the > saved session and then giving user a dialog box to OK the moving of the > windows to their respective desktops, positioning them at the saved location > and resizing them as per the saved info. > > The user just waits for all the application windows to start in the current > desktop and then, pressing OK does the work. There are tricks to center the > dialog box, keep it above all windows and activating it. kdotool works > really well to do the rest. > > Checkout the gist I posted. Most of code can be used as is. Only binary > names need adjusting/addition/deletion. What does kwin-fix_crazy_gesture.patch do? Seems to be a patch to the kwin source code. (In reply to Patrick O'Callaghan from comment #203) > (In reply to devsk from comment #202) > > (In reply to Roke Julian Lockhart Beedell from comment #198) > > > (In reply to imaginator from comment #197) > > > On older systems, this wouldn't really work, since there's always something > > > that randomly takes ages to launch. Is there a way to wait for a specific > > > process to be invoked? If so, perhaps wait for `plasmashell` to be running, > > > then run the script, since it's usually one of the last things to run. > > > https://stackoverflow.com/a/9118509/9731176 might be of use. > > > > I simply went with the option of starting all the applications from the > > saved session and then giving user a dialog box to OK the moving of the > > windows to their respective desktops, positioning them at the saved location > > and resizing them as per the saved info. > > > > The user just waits for all the application windows to start in the current > > desktop and then, pressing OK does the work. There are tricks to center the > > dialog box, keep it above all windows and activating it. kdotool works > > really well to do the rest. > > > > Checkout the gist I posted. Most of code can be used as is. Only binary > > names need adjusting/addition/deletion. > > What does kwin-fix_crazy_gesture.patch do? Seems to be a patch to the kwin > source code. On all platforms where I have used the swipe gesture, it takes to the right desktop when you swipe left. With X11, it was like that (imagine years of motor tuning). With wayland, it drove me crazy because every time I wanted to go a desktop, it took me to the wrong desktop. This behavior would be typically changeable but everything I tried using touchegg, confused the desktop navigation more: touchegg would move it to next but kwin would move it back, meaning that it did not change the desktop. After some real desperation, I took the drastic step of finding and toggling the gesture in code. (In reply to devsk from comment #204) > (In reply to Patrick O'Callaghan from comment #203) > > (In reply to devsk from comment #202) > > > (In reply to Roke Julian Lockhart Beedell from comment #198) > > > > (In reply to imaginator from comment #197) > > > > On older systems, this wouldn't really work, since there's always something > > > > that randomly takes ages to launch. Is there a way to wait for a specific > > > > process to be invoked? If so, perhaps wait for `plasmashell` to be running, > > > > then run the script, since it's usually one of the last things to run. > > > > https://stackoverflow.com/a/9118509/9731176 might be of use. > > > > > > I simply went with the option of starting all the applications from the > > > saved session and then giving user a dialog box to OK the moving of the > > > windows to their respective desktops, positioning them at the saved location > > > and resizing them as per the saved info. > > > > > > The user just waits for all the application windows to start in the current > > > desktop and then, pressing OK does the work. There are tricks to center the > > > dialog box, keep it above all windows and activating it. kdotool works > > > really well to do the rest. > > > > > > Checkout the gist I posted. Most of code can be used as is. Only binary > > > names need adjusting/addition/deletion. > > > > What does kwin-fix_crazy_gesture.patch do? Seems to be a patch to the kwin > > source code. > > On all platforms where I have used the swipe gesture, it takes to the right > desktop when you swipe left. With X11, it was like that (imagine years of > motor tuning). With wayland, it drove me crazy because every time I wanted > to go a desktop, it took me to the wrong desktop. > > This behavior would be typically changeable but everything I tried using > touchegg, confused the desktop navigation more: touchegg would move it to > next but kwin would move it back, meaning that it did not change the > desktop. After some real desperation, I took the drastic step of finding and > toggling the gesture in code. So this patch has actually nothing to do with session restore. I'd suggest posting it somewhere else to avoid confusion. I've had a quick look at your scripts but haven't tried them. They seem quite complicated, but maybe they do more than mine. I've taken note of how you use 'kdotool windowmove' and have added this to my own scripts (don't know why I hadn't noticed this before :-). I'll upload them shortly (still not very familiar with Gist to be honest). (In reply to Patrick O'Callaghan from comment #194) > (In reply to Patrick O'Callaghan from comment #183) > > (In reply to Roke Julian Lockhart Beedell from comment #182) > > > (In reply to Patrick O'Callaghan from comment #181) > > > If you uploaded the script to even something as simple as a GitHub Gist or > > > GitLab Snippet, and then linked that here, I think it'd be more useful, > > > since questions and suggestions can't be posted here without notifying 118 > > > people. Thanks for it, though. > > > > Good idea. I've put it on: > > > > https://gist.github.com/pjoc/a5059b93321ee67d42f4af8d45b7276f > > > > Thanks > > Updated the Restore script to handle window titles with special characters. Updated again to save/restore window positions. (In reply to Patrick O'Callaghan from comment #205) > > So this patch has actually nothing to do with session restore. I'd suggest > posting it somewhere else to avoid confusion. Yes. I don't think anybody cares about flipping the gestures behavior. I will keep this patch around. > I've had a quick look at your scripts but haven't tried them. They seem > quite complicated, but maybe they do more than mine. I've taken note of how > you use 'kdotool windowmove' and have added this to my own scripts (don't > know why I hadn't noticed this before :-). I'll upload them shortly (still > not very familiar with Gist to be honest). Yes. My script does do way more..:-) It saves a lot more information in a format which is easily read in bash. Then, it starts all the apps and waits for the user to ack the restoration process which restores their desktop, position and size. I had to add a lot of code for robustness e.g. some random processes make their way into 'kdotool search' when they don't have a real window associated with them e.g. plasmashell. For example, restoring from X11 to wayland, the window class changes from konsole to org.kde.konsole. For example, the size and location on wayland are floating point numbers sometimes, which was not apparent until wayland locked up on creating a window with 0 size with infinite memory allocation loop and got killed by OOM. For example, if you saved session from a crontab periodically, kdotool just doesn't find any windows because crontab does not have access to the user environment to access the DBUS. Lots of accounting for quirks. (In reply to devsk from comment #207) > (In reply to Patrick O'Callaghan from comment #205) > > > > So this patch has actually nothing to do with session restore. I'd suggest > > posting it somewhere else to avoid confusion. > > Yes. I don't think anybody cares about flipping the gestures behavior. I > will keep this patch around. I'd suggest keeping it separate from the rest. It's confusing to have it lumped in with the save/restore stuff. > > I've had a quick look at your scripts but haven't tried them. They seem > > quite complicated, but maybe they do more than mine. I've taken note of how > > you use 'kdotool windowmove' and have added this to my own scripts (don't > > know why I hadn't noticed this before :-). I'll upload them shortly (still > > not very familiar with Gist to be honest). > > Yes. My script does do way more..:-) It saves a lot more information in a > format which is easily read in bash. Then, it starts all the apps and waits > for the user to ack the restoration process which restores their desktop, > position and size. All the apps I've used already remember their sizes on restarting, so I don't bother with that. I'll consider using Zenity or similar, but given that it only helps with restoring and not saving, I wouldn't place a high priority on it. If there were a reliable way to save state on logging out, that would change. Unfortunately the Auto-start logout scripts only seem to trigger once all windows have been destroyed. > I had to add a lot of code for robustness e.g. some random processes make > their way into 'kdotool search' when they don't have a real window > associated with them e.g. plasmashell. I already skip plasmashell, and if anything else has this behaviour it would be trivial to add. > For example, restoring from X11 to wayland, the window class changes from > konsole to org.kde.konsole. I don't care about X11. I wrote this specifically for Wayland. > For example, the size and location on wayland are floating point numbers > sometimes, which was not apparent until wayland locked up on creating a > window with 0 size with infinite memory allocation loop and got killed by > OOM. > kdotool can return floating-point values for position, but doesn't accept them when trying to set it. This to me is a bug. However I'm not worried about fractions of a pixel, so I just truncate these to integers. > For example, if you saved session from a crontab periodically, kdotool just > doesn't find any windows because crontab does not have access to the user > environment to access the DBUS. Not something I do. I only save the state explicitly. > All the apps I've used already remember their sizes on restarting, so I don't bother with that.
konsole doesn't do that for me. It opens a single window of a dimension from the profile.
My script opens as many konsole windows as you had open before, moves to the right location in the right desktop and sizes them.
(In reply to devsk from comment #209) > > All the apps I've used already remember their sizes on restarting, so I don't bother with that. > > konsole doesn't do that for me. It opens a single window of a dimension from > the profile. > > My script opens as many konsole windows as you had open before, moves to the > right location in the right desktop and sizes them. Thanks for pointing that out I almost never have more than one Konsole terminal open so hadn't noticed. It does however remember the size of the single one it opens. Can someone please give a status update on Wayland's progress on the protocol mentioned in Comment #89? https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18 I believe this is the protocol of the topic. Turned 5 years now. Hurray, happy anniversary! :-/ Some coverage: https://www.youtube.com/watch?v=UAMAlqqVijY (In reply to Eugene Savitsky from comment #212) > https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18 > > I believe this is the protocol of the topic. Turned 5 years now. Hurray, > happy anniversary! :-/ What are 5 years in the ocean of time. :) (In reply to Kevin Gilbert from comment #211) > Can someone please give a status update on Wayland's progress on the > protocol mentioned in Comment #89? That can only happen if someone was working on it. Nobody is. The presence of this bug gives that wrong impression. I gave up on this bug ever getting fixed. That's why I ended up writing the set of elaborate shell functions I wrote. They do most of the work for session restore for me and I have been super happy with that. It works better than both the x11 and wayland restore in KDE right now. This is a bug tracker, not a forum. If you're not adding something, don't post things. People are working on this. There's the window management stuff in kwin that's linked somewhere, rebased very recently and there's also Qt work here: https://codereview.qt-project.org/c/qt/qtwayland/+/603992 But that's not the important part of session restore. You need to actually restart the apps. We cannot blindly copy XSMP because that's a huge security hole for Flatpak'd apps. *including on X11* It is going to need something that that allows apps in flatpaks to request a restart command and have something outside that rewrite that command with the security prefixing. This is the part that's in progress in xdg-desktop-portals and why the Wayland part is seemingly stuck. (In reply to David Edmundson from comment #216) > This is a bug tracker, not a forum. If you're not adding something, don't > post things. I posted something that gets me 95% of the way there. If KDE had gotten me there 1/2/3/4 years ago, I would have moved to wayland much earlier and never complained. > People are working on this. Why so secret about it if people are working on it? We rarely see any updates here or anywhere else I could find with google search. What I could find showed that the development stopped more than 2 years ago. > This is the part that's in progress in xdg-desktop-portals and why the > Wayland part is seemingly stuck. Can you please point to the discussion thread for this? I need read-only access and have no (or had any) intention of spamming anyone. The last thing I want to do is to annoy the people who are working voluntarily on this feature. (In reply to devsk from comment #217) > (In reply to David Edmundson from comment #216) > > This is a bug tracker, not a forum. If you're not adding something, don't > > post things. > > > People are working on this. > > Why so secret about it if people are working on it? We rarely see any > updates here or anywhere else I could find with google search. What I could > find showed that the development stopped more than 2 years ago. Seems not stopped, and still have some activity and progress: https://www.youtube.com/watch?v=UAMAlqqVijY https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18#note_2657667 However not clear: roadmap, blockers, estimate, etc. Open source projects sometime lacks such info. > > This is the part that's in progress in xdg-desktop-portals and why the > > Wayland part is seemingly stuck. Any wiki, forum, info where we can track progress. problems? Modern tendencies forces us to switch to wayland, however many important features are still missed and a lot of Known Significant Issues exists: https://community.kde.org/Plasma/Wayland_Known_Significant_Issues (In reply to Aleksey Kontsevich from comment #218) > However not clear: roadmap, blockers, estimate, etc. Open source projects sometime lacks such info. As you've just stated, https://invent.kde.org/plasma/kwin/-/milestones would be usable if there were milestones for this. If you want it to be used, I'd suggest you posit it in Discourse in the Brainstorm category, because that, if it *is* a problem, evidently affects more than solely this ticket. > Any wiki, forum, info where we can track progress. problems? If this issue really isn't enough, you can watch the MediaWiki page at https://community.kde.org/index.php?title=Plasma/Wayland_Known_Significant_Issues&oldid=103027#Waiting_on_new_Wayland_protocol:~:text=Upstream%5Bedit%5D,Hibernation%20(Suspend%20To%20Disk)%20instead. https://community.kde.org/index.php?title=Plasma/Next/SessionManagement&oldid=35103#Integration_Story_2:~:text=Started%20by%20the%20startkde%20script%2C,kwin%20in%20kwin's%20session%20file *might* work too, but I'd doubt that it'll ever change. Please just stop spamming for updates, everyone who is. You'll get them when you see a commit comment from the Bug Janitor Service. (In reply to Roke Julian Lockhart Beedell from comment #219) > (In reply to Aleksey Kontsevich from comment #218) > > However not clear: roadmap, blockers, estimate, etc. Open source projects sometime lacks such info. > > As you've just stated, https://invent.kde.org/plasma/kwin/-/milestones would > be usable if there were milestones for this. If you want it to be used, I'd > suggest you posit it in Discourse in the Brainstorm category, because that, > if it *is* a problem, evidently affects more than solely this ticket. > > > Any wiki, forum, info where we can track progress. problems? > > If this issue really isn't enough, you can watch the MediaWiki page at > https://community.kde.org/index.php?title=Plasma/ > Wayland_Known_Significant_Issues&oldid=103027#Waiting_on_new_Wayland_protocol > :~:text=Upstream%5Bedit%5D,Hibernation%20(Suspend%20To%20Disk)%20instead. > https://community.kde.org/index.php?title=Plasma/Next/ > SessionManagement&oldid=35103#Integration_Story_2:~: > text=Started%20by%20the%20startkde%20script%2C, > kwin%20in%20kwin's%20session%20file *might* work too, but I'd doubt that > it'll ever change. > > Please just stop spamming for updates, everyone who is. You'll get them when > you see a commit comment from the Bug Janitor Service. You don't seem to realize the gravity of the situation. The sheer number of people on cc should give you an idea. But I guarantee you that this is only the tip of the iceberg. This bug is critical and several years old already. It's a real strain on users' nerves. There's no noticeable progress. And other than this thread practically no information. No wonder patience is coming to an end. Thus, instead of brushing off legitimate requests for information as *spamming* and expecting users to just shut up, sit still and wait in awe for comments from the Bug Janitor Service, I suggest you either significantly increase your efforts to find a solution for this or at least re-evaluate your communication policy and your attitude towards yet loyal users. If you don't want to go under, that is. Thanks. (In reply to imaginator from comment #220) > I suggest you either significantly increase your efforts to find a solution for this or at least re-evaluate your communication policy and your attitude towards yet loyal users. I don't represent KDE. Where did you get that from?! I just don't want my mailbox filled with whining. Asking for updates here, no matter how much it annoys any of us, provides 0 help to the developers working on it. The relevant FreeDesktop and KDE GitLab URIs have already been provided. If you register accounts there and subscribe to the threads, you'll receive updates. I get them every few months when a few new commits are pushed. To summarise, you can subscribe to: 1. https://community.kde.org/index.php?title=Plasma/Wayland_Known_Significant_Issues 2. https://bugs.kde.org/show_bug.cgi?id=436318#c0 3. https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18 4. https://invent.kde.org/plasma/kwin/-/issues/113 Certainly, continue to post greivances here, if you wish (they're much better put here than there), but try to *add something*. This is an >200-comment thread, so most things you can think of have already been said. (In reply to Roke Julian Lockhart Beedell from comment #221) > (In reply to imaginator from comment #220) > > I suggest you either significantly increase your efforts to find a solution for this or at least re-evaluate your communication policy and your attitude towards yet loyal users. > > I don't represent KDE. Where did you get that from?! I just don't want my I inferred that from the kind of and wording of your reply to a user who was simply asking for information. > mailbox filled with whining. Asking for updates here, no matter how much it > annoys any of us, provides 0 help to the developers working on it. Who's "whining" here? Most are just seeking information that is hard to find elsewhere. Some opposed attempts to prematurely close this bug. Others contributed (partial) workarounds. I can't see anything wrong in that. Indeed, I found that more instructive and constructive than the sound of silence from the "officials". > > The relevant FreeDesktop and KDE GitLab URIs have already been provided. If > you register accounts there and subscribe to the threads, you'll receive > updates. I get them every few months when a few new commits are pushed. To > summarise, you can subscribe to: > > 1. > https://community.kde.org/index.php?title=Plasma/ > Wayland_Known_Significant_Issues > 2. https://bugs.kde.org/show_bug.cgi?id=436318#c0 > 3. > https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18 > 4. https://invent.kde.org/plasma/kwin/-/issues/113 > > Certainly, continue to post greivances here, if you wish (they're much > better put here than there), but try to *add something*. This is an > >200-comment thread, so most things you can think of have already been said. Again, I don't like your deprecating wording and I find it completely unfitting. Contributions to deal with this mess have been made, by me and others and if you can't rate that as "something added", so be it. But one cannot expect the users here to solve a bug that so far has been unsolvable for the experts. And BTW: what have _you_ added to solve this? I consider my reply to your post indeed a contribution as it hopefully brought it home to the KDE-guys that this is a _huge_ problem and that they urgently need to do something about it or at least improve their communication. Just hushing this up is not going to work much longer. But if you don't like being bothered by "whinings" or "grievances" like that, simply take yourself off the cc-list. (In reply to imaginator from comment #222) You may be correct, but this isn't the place to discuss grievances with a specific person. For the sake of everyone else's mailbox, if you want to discuss this, PM me on Discourse. I posted my comment to consolidate the relevant tracking URIs into a single comment. David Edmundson is it resolved? What Plasma version? How to test? Is it the same session restore like in X11, or better, or some fake restore? Could You explain please? Thanks! What does a "Status: Closed Later" mean? Maybe the resolution description is too broad. Perhaps if the problem is broken down into smaller pieces some of he users would be satisfied withe some of the steps toward the final objective. For me the first piece would be to restore the windows to the proper desktop(s). I realize there are subsets to the solution like same window on multiple desktops, of multiple instances of the application in the same window. But first define the end objective and prioritize the steps to get there. As a part of this does anyone know what is the format of the session save file? To restore you first have to save the current information correctly. > David Edmundson is it resolved? > What does a "Status: Closed Later" mean? Now that you guys have pissed off the main developer, whatever implementation was coming in the spring is now off the table. just kidding....:-) Nothing was coming this spring. All David had to do was to remove himself from the CC list and he would be fine. But he had to use his developer hammer. The patience and the kindness comes difficult these days! In the meantime, I made a change to my code snippet to match the title with regex because I realized that if the title of the window had some information that was variable (e.g. a tab in firefox has yahoo mail and N unread messages, and next moment it is N+10), it won't restore it correctly. In case you want a poor man's session restore: https://gist.github.com/devsk/9eff2d08d6cd5772afd2b178930b66ae PS: This feature will require a fresh, young (think college grad) set of eyes and will come in due time, and when it comes, it will come fast! I am very hopeful. With https://invent.kde.org/plasma/kwin/-/merge_requests/7475, preliminary support for session restore has been merged for Plasma 6.4! This implements the WIP wayland protocol at https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18. Since the protocol itself is WIP, things could change, and support should be considered somewhat experimental at the moment. Also keep in mind that since it's a new protocol, toolkits and apps need to opt into it. So don't expect everything to immediately start working as soon as you upgrade to Plasma 6.4. Support from apps should start trickling in over time, likely with KDE apps leading the way. There's still an open question as to what we do with apps that haven't opted in yet. This is still being discussed. Regardless, things are moving! Thanks for your patience here, everyone. Soon our long national nightmare of no session restore on Wayland will be over. :) (In reply to Nate Graham from comment #228) > With https://invent.kde.org/plasma/kwin/-/merge_requests/7475, preliminary > support for session restore has been merged for Plasma 6.4! > > This implements the WIP wayland protocol at > https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18. > Since the protocol itself is WIP, things could change, and support should be > considered somewhat experimental at the moment. > > Also keep in mind that since it's a new protocol, toolkits and apps need to > opt into it. So don't expect everything to immediately start working as soon > as you upgrade to Plasma 6.4. Support from apps should start trickling in > over time, likely with KDE apps leading the way. There's still an open > question as to what we do with apps that haven't opted in yet. This is still > being discussed. > > Regardless, things are moving! Thanks for your patience here, everyone. Soon > our long national nightmare of no session restore on Wayland will be over. :) That's encouraging news. Does this mean that at least window placement on desktops will start working, even if individual windows don't restore their state? (In reply to Patrick O'Callaghan from comment #229) > (In reply to Nate Graham from comment #228) > > With https://invent.kde.org/plasma/kwin/-/merge_requests/7475, preliminary > > support for session restore has been merged for Plasma 6.4! > > > > This implements the WIP wayland protocol at > > https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18. > > Since the protocol itself is WIP, things could change, and support should be > > considered somewhat experimental at the moment. > > > > Also keep in mind that since it's a new protocol, toolkits and apps need to > > opt into it. So don't expect everything to immediately start working as soon > > as you upgrade to Plasma 6.4. Support from apps should start trickling in > > over time, likely with KDE apps leading the way. There's still an open > > question as to what we do with apps that haven't opted in yet. This is still > > being discussed. > > > > Regardless, things are moving! Thanks for your patience here, everyone. Soon > > our long national nightmare of no session restore on Wayland will be over. :) > > That's encouraging news. Does this mean that at least window placement on > desktops will start working, even if individual windows don't restore their > state? Properties like desktop number, window position and window size should all be independent of the apps opting in. kwin controls these. This is 95% of the way there. Its the internal state of the app, like how many tabs of konsole in a window on desktop 3 or which file in kate on desktop 1 was open last, that will take some work on the app side. Nate, is my understanding correct? PS: Thanks for this much needed update! And thanks to David for working on the PR. Appreciate your efforts! > Properties like desktop number, window position and window size should all be independent of the apps opting in. kwin controls these. This is 95% of the way there.
Does kwin also control which activity a windows was placed in?
To clarify: - The protocol is implemented; support for it needs to be added app-by-app. - The protocol currently supports window geometry, positioning, and virtual desktop but not actual in-app content. I don't know of the status for Activities. - In-app content restoration is also planned, and will come later. Thanks for the update Nate! (In reply to Nate Graham from comment #232) > - The protocol is implemented; support for it needs to be added app-by-app. If there is a Wiki page, which enlists all apps and their current state of implementation (fully supported, partially supported, planned), it would be great to link it here, otherwise I'd make sense to create one I'd say. It's a bit more complicated than that. We can probably opt all KDE apps into the window sizing and positioning stuff all at once by changing something in they all use. Individual KDE apps that do have additional state to save will need to opt into saving it one by one, and also only after the Wayland protocol supports this. 3rd-party apps are in a similar boats; for example GNOME may also be able to opt all their apps in all at once, but would need individual apps to opt into saving their app-specific states. Non-GNOME GTK apps will likely need to be opted into each individual thing one by one. The situation with other toolkits like Electron isn't clear to me. A wiki page with all this information sounds like a massive coordination project to undertake. Also it's not really KDE-specific, so it should probably be done somewhere upstream. Personally I don't have the bandwidth for a project like, so someone else would need to give that a go. If anyone else wants to work on it, please feel free! There's a lot of confusion still. The protocol on it's own does not help with apps restoring their position size or desktop *even if* the app opted into this protocol. It would still need to know it's restoring an older state which means we it goes hand-in-hand with the in-app content restoration protocol. It it not a user-facing useful step on it's own. (In reply to Nate Graham from comment #228) > With https://invent.kde.org/plasma/kwin/-/merge_requests/7475, preliminary > support for session restore has been merged for Plasma 6.4! > > This implements the WIP wayland protocol at > https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/18. > Since the protocol itself is WIP, things could change, and support should be > considered somewhat experimental at the moment. > > Also keep in mind that since it's a new protocol, toolkits and apps need to > opt into it. So don't expect everything to immediately start working as soon > as you upgrade to Plasma 6.4. Support from apps should start trickling in > over time, likely with KDE apps leading the way. There's still an open > question as to what we do with apps that haven't opted in yet. This is still > being discussed. > > Regardless, things are moving! Thanks for your patience here, everyone. Soon > our long national nightmare of no session restore on Wayland will be over. :) Is this actually usable now? In the various announcements for Plasma 6.4 I haven't noticed any mention of it. *** Bug 503259 has been marked as a duplicate of this bug. *** With https://codereview.qt-project.org/c/qt/qtbase/+/649886, support has landed in Qt 6.10 for real Wayland session restore. This means that KDE apps can start opting in, and then they can benefit from *real* session restore when built against Qt 6.10. To be clear: nothing user-facing will visibly change until KDE apps opt in and are running on top of Qt 6.10. Non-KDE apps will also be able to open in once they or their UI toolkit implements support for the xx-session-management-v1protocol. (In reply to Nate Graham from comment #238) > With https://codereview.qt-project.org/c/qt/qtbase/+/649886, support has > landed in Qt 6.10 for real Wayland session restore. > > This means that KDE apps can start opting in, and then they can benefit from > *real* session restore when built against Qt 6.10. > > To be clear: nothing user-facing will visibly change until KDE apps opt in > and are running on top of Qt 6.10. > > Non-KDE apps will also be able to open in once they or their UI toolkit > implements support for the xx-session-management-v1protocol. To clarify: does this mean that only apps which have opted in will have their desktop and window positions restored? Not talking about window contents, which does of course req (In reply to Patrick O'Callaghan from comment #239) > (In reply to Nate Graham from comment #238) > > With https://codereview.qt-project.org/c/qt/qtbase/+/649886, support has > > landed in Qt 6.10 for real Wayland session restore. > > > > This means that KDE apps can start opting in, and then they can benefit from > > *real* session restore when built against Qt 6.10. > > > > To be clear: nothing user-facing will visibly change until KDE apps opt in > > and are running on top of Qt 6.10. > > > > Non-KDE apps will also be able to open in once they or their UI toolkit > > implements support for the xx-session-management-v1protocol. > > To clarify: does this mean that only apps which have opted in will have > their desktop and window positions restored? Not talking about window > contents, which does of course req Yep, not clear: under X11 all app are restored regardless they support session management or not. (In reply to Aleksey Kontsevich from comment #240) > Yep, not clear: under X11 all app are restored regardless they support > session management or not. What do you find unclear here? In X11, it is the X11 session management which is responsible of keeping track of which applications belong to a session. In Wayland, it is up to the application to remember that it belongs to a session. (In reply to Lassi Väätämöinen from comment #241) > (In reply to Aleksey Kontsevich from comment #240) > > > Yep, not clear: under X11 all app are restored regardless they support > > session management or not. > > What do you find unclear here? In X11, it is the X11 session management > which is responsible of keeping track of which applications belong to a > session. > > In Wayland, it is up to the application to remember that it belongs to a > session. I do not much care how Wayland/X11 algorithm works, I'm curious if I get the same result/behavior in Wayland comparing to X11. Seems not! So do not push us to migrate to weird buggy Wayalnd. (In reply to Lassi Väätämöinen from comment #241) > (In reply to Aleksey Kontsevich from comment #240) > > > Yep, not clear: under X11 all app are restored regardless they support > > session management or not. > > What do you find unclear here? In X11, it is the X11 session management > which is responsible of keeping track of which applications belong to a > session. > > In Wayland, it is up to the application to remember that it belongs to a > session. I for one didn't know that. I like to think I'm a fairly knowledgeable user, but I'm not a GUI programmer. I understand that apps need to take of their own internal state, but their desktop and window placement are presumably not their concern. Maybe I'm wrong or we're using the terminology in contradictory ways. What I do know is that with kdotool I can place and resize windows from an external script, so I fail to see why the desktop manager can't do this for me. First of all, a really thanks to the developers for pushing this forward. Really appreciated. One critical question: wouldnt it make sense to keep this bug open until at least Qt 6.10 is finally released? From what i see the planned date for the final release is planned for the 23rd of September. So apart from any development system, this fix will actually reach no users before that date. (In reply to ilovekiruna from comment #244) > One critical question: wouldnt it make sense to keep > this bug open until at least Qt 6.10 is finally released? Depending on how an organizations bug process is defined, the answer is maybe yes or maybe no. From KDE perspective, the bug is fixed. It's the upstream that this depends on. So perhaps the upstream issue could be linked here. >this fix will actually reach no users before that date. What comes to this part, the real user-visible fix(es) will only show once the applications adopt this, which may be now, later, or never. So in that regard it doesn't make sense to keep this big open, as far as I see it. (In reply to Lassi Väätämöinen from comment #245) > From KDE perspective, the bug is fixed. It's the upstream that this depends > on. So perhaps the upstream issue could be linked here. > > >this fix will actually reach no users before that date. > > What comes to this part, the real user-visible fix(es) will only show once > the applications adopt this, which may be now, later, or never. So in that > regard it doesn't make sense to keep this big open, as far as I see it. Yep, this is accurate and reflects the Bugzilla ticket lifecycle that KDE uses. The next step is for client software (Plasma and apps) to implement support now. The big problem is that the true solution needs to be done multiple times. That is at least once for each application instead of once for each session manager. I agree that each application needs to keep track of its own state(s) but the solution provided is really a step back. That said hopefully there will be some sample applications which will make it easier to implement the new protocol. I also suspect there may be situations where one might not want to restore the application or the application state. (In reply to Brian Kaye from comment #247) > I also suspect there may be situations where one might not want to restore the application or the application state. I agree on this point. I've, many a time, had this be a problem on AOSP and Windows 11. Would this be implementable in KWin, or require standardisation by FreeDesktop? I'd presume merely the former. (In reply to Brian Kaye from comment #247) > The big problem is that the true solution needs to be done multiple times. > That is at least once for each application instead of once for each session > manager. I agree that each application needs to keep track of its own > state(s) but the solution provided is really a step back. +1 ! Absolutely!!! I would prefer solution where session manager takes care of this: at least just launch an app and restores its position, size and state (minimized, maximized, minimized to tray, etc). If application want and it is necessary it may implement more comprehensive inner state restore using Wayland protocol, but main functionality should not require this. (In reply to Aleksey Kontsevich from comment #249) > (In reply to Brian Kaye from comment #247) > > The big problem is that the true solution needs to be done multiple times. > > That is at least once for each application instead of once for each session > > manager. I agree that each application needs to keep track of its own > > state(s) but the solution provided is really a step back. > > +1 ! Absolutely!!! > > I would prefer solution where session manager takes care of this: at least > just launch an app and restores its position, size and state (minimized, > maximized, minimized to tray, etc). If application want and it is necessary > it may implement more comprehensive inner state restore using Wayland > protocol, but main functionality should not require this. +1 That's essentially what I've been trying to say. Some applications I use (eg Dolphin, Firefox/Waterfox) already seem to restore their state independent of Wayland. I still do not understand that since X11 solved the session restore problem, why could that solution not be imported into Wayland or whatever the session restore process is now. If memory serves me, there was an entry made by the session manager on logout for each application to ~.config/session ( still seems to be used for dolphin and kwin). (In reply to Brian Kaye from comment #251) > I still do not understand that since X11 solved the session restore problem, > why could that solution not be imported into Wayland or whatever the session > restore process is now. I guess the short answer is something in the lines of: because Wayland was not meant to be the same bloatware as X11 is. And adding all the features in it would make it as inflexible as X11 was. But as Nate has already mentioned (I guess several times), the bug comment section is really about techincal discussion regarding the actual bug, not general. So I'd suggest moving the general dicussion to https://discuss.kde.org/. (In reply to Lassi Väätämöinen from comment #252) > (In reply to Brian Kaye from comment #251) > > I still do not understand that since X11 solved the session restore problem, > > why could that solution not be imported into Wayland or whatever the session > > restore process is now. > > I guess the short answer is something in the lines of: because Wayland was > not meant to be the same bloatware as X11 is. And adding all the features in > it would make it as inflexible as X11 was. To compare one of the most productivity-enhancing features of modern desktops with "bloat" is somewhat beside the point, I'd say. And implementing such a key-feature surely wouldn't have made wayland "inflexible". That's ridiculous. Thus, another "short answer" might be: the Wayland-devs didn't care or weren't up to the task - despite a blueprint (XSMP) that's been around for more than 30 years. And BTW: if I were a QT- or KDE-developer, I'd be _very_ cautious to point fingers at others when it comes to "bloatware". I hope you're wise enough to refrain from asking for hard data. > > But as Nate has already mentioned (I guess several times), the bug comment > section is really about techincal discussion regarding the actual bug, not > general. So I'd suggest moving the general dicussion to > https://discuss.kde.org/. Had KDE-, QT-, or wayland-devs solved this problem after 4 weeks or 4 months or even a year, there wouldn't be a discussion at all. But after more than _4 years_ I guess some perhaps not so technical - but nevertheless mostly constructive - comments are quite understandable. And in case you have opened a thread for this topic on https://discuss.kde.org/: could you post the link, please? Thanks. Anyway - it's good that there is progress in this matter and that we've been informed about it. Thanks. (In reply to imaginator from comment #253) Well, here you go, here's one touching the topic, for example, on a quick search: https://discuss.kde.org/t/this-week-in-plasma-the-beginnings-of-wayland-session-restore-kde-blogs/32728 What to discuss there?! You know our opinion, no sense to repeat it there as you'll ignore it same way like here, no sense to waste our time and effort. (In reply to Aleksey Kontsevich from comment #255) Asking that merely invites discussion here. Ask that at the linked Discussion thread. (In reply to Nate Graham from comment #246) > (In reply to Lassi Väätämöinen from comment #245) > > From KDE perspective, the bug is fixed. It's the upstream that this depends > > on. So perhaps the upstream issue could be linked here. > > > > >this fix will actually reach no users before that date. > > > > What comes to this part, the real user-visible fix(es) will only show once > > the applications adopt this, which may be now, later, or never. So in that > > regard it doesn't make sense to keep this big open, as far as I see it. > > Yep, this is accurate and reflects the Bugzilla ticket lifecycle that KDE > uses. > > The next step is for client software (Plasma and apps) to implement support > now. So -- just to clarify -- are app-specific instances of this bug (e.g. https://bugs.kde.org/show_bug.cgi?id=461780 for konsole) supposed to stay open, or are they to be marked as duplicates of this bug (as has been done, it seems, in its earlier history)? Sorry, I post here as I can't log into the discussion thread page. On 6.3.6, under Wayland, I can set to save the session and Dolphin is restored at startup, so it remains to implement session restore in Dolphin to set it up as the last time (tabs, URIs and so on). Question is: almost every common Plasma application throws a message at restart/shutdown time asking whether to close it. For Dolphin, under X11, the message is ignored and the restart/shutdown process is continued, but for Yakuake is blocker, so the process remains waiting for the user confirmation indefinitely. Under Wayland it is a blocker for Dolphin too (issuing a reboot command in command line reboots the system and the previous session is restored). If I click quit, I am not sure of the result at session restore. This feature stays in the way somehow and creates a lot of confusion. *** Bug 509869 has been marked as a duplicate of this bug. *** |