| Summary: | Latest KMM Master Version is aborting at 29% of the build | ||
|---|---|---|---|
| Product: | [Applications] kmymoney | Reporter: | Michael Carpino <mfcarpino> |
| Component: | buildsystem | Assignee: | KMyMoney Devel Mailing List <kmymoney-devel> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | NOR | ||
| Version: | git (master) | ||
| Target Milestone: | --- | ||
| Platform: | Compiled Sources | ||
| OS: | Linux | ||
| Latest Commit: | 604dae022e9544e34e0c411ca1d79f29fbd6f049 | Version Fixed In: | |
| Sentry Crash Report: | |||
| Attachments: | Output regarding abort | ||
Please specify the version of qt you have, I see cmake (3.11), gcc (8.1.1), qt, kde appllications (5.46.0), and kde-frameworks (5.8.2) from the output you attached.) I have been having this problem for some time, but nobody else has been unable to replicate it, so any evidence that will help figure out the actual cause will be helpful. (I'm on Artix LInux, and I think all my versions are the same or higher, with qt 5.11) I still suspect some gcc difference, but have no real proof. I would suspect qt, but I'm pretty sure we differ by minor version only, and they don't break API compatability within major version. I'm running qt version 5.11.0-1. Let me know if there is anything else. I updated KMM fine last weekend and then I received probably 200 updates for various programs including the latest kernel yesterday and today. Then I manually tried to update KMM and then the abort happened. This is the part that makes me wonder: no known conversion for argument 1 from ‘QHeaderView*’ to ‘const QObject*’ I agree with Thomas in Comment 3 that this particular error just doesn't really make sense, as following that arguement all the way through the templates and macros (assuming I've done it correctly) it does seem the correct type. I marked the bug as Confirmed, since it's showed up more than once, and changed from Manjaro to compiled sources, as this is not depndent on distribution. Also - I just got the same error compiling with craft on Windows 10. It said [727/995] so I assume it was nina and not make, although it shouldn't matter. What is odd (from my perspective is that gcc is 7.3.0, so gcc 8 is not required for this error. Frameworks are 5.47 and qt is 5.11 so I'm now thinking it is most likely one of them - although I don't think either is supposed to break API compatability. I wonder if this is worth posting to some QT mailing list or forum? Any other ideas? The patch just committed by Jonathan Riddell seems to have fixed this, at least on Artix Linux. I will test later with craft on Win 10. My guess now is that between QT 5.9 and QT 5.11, "#include <QHeaderView>" was dropped from some other header file where it had indirectly been included. Now we need to explicitly include it. Thanks for the update, Jack. The patch allowed me to complete the install as well. I'll leave the Bug open for a few weeks to see if anyone still has any issues. If nobody comments I'll then close the Bug. Interestingly, on Artix Linux I actually needed one more patch to complete the build - adding "#include <QAbstractItemView>" to kmymoney/views/newtransactioneditor.cpp. The craft build is running now. Git commit 61fff98b33b2127c8a07e5c41b8b1e962f501676 by Thomas Baumgart. Committed on 22/06/2018 at 14:10. Pushed by tbaumgart into branch 'master'. Fix compilation with Qt5.11 M +1 -0 kmymoney/views/newtransactioneditor.cpp https://commits.kde.org/kmymoney/61fff98b33b2127c8a07e5c41b8b1e962f501676 |
Created attachment 113206 [details] Output regarding abort I'm on Manjaro RELEASE=17.1.10 with Kernel 4.17.0-2 and the build is aborting, See the attachment for the output produced with the abort.