Version: (using KDE 4.3.0) Installed from: SuSE RPMs On one hand user sees: This crash information is not useful and on the other (when tries to quit) Do you really want to close the bug assistant without submitting the bug report? Please check first, if such question makes sense.
(In reply to comment #0) > Version: (using KDE 4.3.0) > Installed from: SuSE RPMs > > On one hand user sees: > This crash information is not useful > In which step: ? a) backtrace fetching b) analysis results If you are on a): even with a not-100%useful backtrace, the report can be useful in some cases. If you are on b): if the whole report is not worth reporting, this message should not appear. If you are in any other step, the question could be evaluated. Thanks
When backtrace is shown. If it is (a) step then I don't understand your comment (here). Because either is it _not_ useful or it is useful. You cannot send both signals at the same time, because user will be confused. I see zero points, I see clear message the backtrace is useless, ok, let's trash it. If user should send the backtrace anyway, please change the message then (in the dialog).
Mh, ok, I see the point. However, what I wanted to say before was: Even when a backtrace could not be useful (and the message "crash info is not useful" is shown), the whole "report" could be useful if the user provide other useful information like good details or a way to reproduce. "Backtrace usefulness != report usefulness" There are some confusions here; so may be the text "the crash information is not useful" could be renamed to "the crash information(or backtrace?) is not useful but the report could be useful" (this is even more confusing, I need a better phrase, but I hope you get the idea) As a possible solution to this confusion, in KDE4.4 I already changed the workflow: before fetching the backtrace, the questions are asked; so, if the user can't provide details or if he is not willing to help, the backtrace step is skipped; and the bug is not reported. This simplify the UI a bit. I agree about the confusing messages, do you see any other way to improve this ? Thanks
I think the reversed steps are regression -- it is counter productive for power users who know the drill and they are want just to get the backtrace. Example case: + random crash (hard to reproduce) + no backtrace I know about randomness already so I don't need any questions, I need backtrace. But back to the questions -- if there is no backtrace there is no danger user quits the dialog, because there is no loss. So confirmation should not be shown. The only chance such report would be OK, if the crash occurs often easily, but in such case, user will encounter it again, and besides, all he/she has to do it is describing the steps. So I would suggest hiding the textarea with backtrace with button "show" (for advanced users) and changing the message: "the data gathered at crash site are too ambiguous. If this crash happened to you second time please proceed and describe the steps how we can repeat it". Something like this. And no confirmation in such case.
If a advanced user/dev only wants to get the backtrace, he/she should use the"Developer information" tab in the main dialog and do not relay on the "Report Bug" feature at all.. In 4.3 series, the Report Bug function will close the main dialog, causing some problems. I'm wanting to change this behaviour with this proposal http://docs.google.com/View?id=dgptsv7v_854nmvbf7 (not reviewed nor implemented yet). Thanks
Going back to the original problem as the feature I described in my last comment is going to be delayed until KDE SC 4.5 (in any case I polised the reporting dialog a bit) The workflow of the dialog changed a bit: The first input page will ask the reporter how much information can be provided...), if this information *could be* useful, the backtrace is fetched. If the backtrace is not useful (0 stars), the user will get a special message telling him/her to install the missing debug packages (a button to do this automatically is over there too...)... and reload the backtrace. So, even if the user gets a zero backtrace in the first attempt, a 100% backtrace can be achieved on a second attempt.... (this is also described in the guide/help) That is why I need to ask the user if he/she wants to close dialog even on that step... THe user needs to *wait* until the conclusions page which will mixup and analyze *all the information* to give a conclusion "Useful" "Not useful" ... This is not going to change... - How would you modify the texts to "avoid this "confusion"" ? (for KDE SC 4.5)
There's really no reason to change this. As Dario mentioned report quality may be improved as you go through the wizard, and in any event since the user can't retrieve data after closing it's always worthwhile to make sure this is the intended action.