I got my Kirigami app to load by patching kirigami.{cpp,h}:

Thanks for this.

<TeleFuchs> on IRC pointed out that this could affect telephone numbers.

I'm afraid this was a breaking API change -- it changes the behaviour, and not all numbers are intended to have separators added in this way.

As the person who raised the Debian Stretch issue last time, I don't think it's worth caring about now that Buster is released.

Nice work!

On further consideration, I think I was wrong about this, sorry.

I'm opposed to the concept of this patch.

Please no.

It's quite easy to accidentally start repeated configure jobs, in which case the second one fails - e.g. by clicking "Apply" and then "Ok" in the project settings.

@rjvbb : If you generate a compile_commands.json manually or with some tool, the best option for KDevelop (as I mentioned in that email) is to import it as a "CMake" project.

+1, as on IRC.

Debian Stretch (current stable release) has Qt 5.7 and KF 5.28. Next stable release isn't until mid-2019.

A year seems long enough. The code's in git if anyone wants it, there's no point advertising a feature that's unusable and not maintained.

The QSet include is redundant now.

This looks good to me overall, but see a few inline comments.

This version deadlocks KDevelop immediately when opening a Python project. Several threads are waiting on cacheMutex.

This is definitely nicer; I remember being a bit puzzled by Helper being a class, and the inscrutable static-var problems are a nuisance.

It may be that it's simply missed a couple of years of VCS API changes, and nothing's fundamentally broken. Even in that case, it would need someone vaguely familiar with CVS to test it properly.

I've tried that (and I believe @kfunk did also).

The test isn't being built or executed at all currently.

Hm, there's a test, which surely ought to fail if the plugin isn't loaded, but doesn't appear in the CI's list of failing tests.

Thanks! Works here.

This might also fix https://bugs.kde.org/show_bug.cgi?id=372686 ?

@arichardson had a previous patch on similar lines: https://git.reviewboard.kde.org/r/126156/

Yeah, I'd find this useful in principle. I don't always remember exactly when I last built/installed my system copy, which can be inconvenient when looking to see which recent commits might be responsible for a problem.

That problem exists with the current 5.2 branch (encountered by me, confirmed by frinring), so it's probably not caused by your patch.

https://cgit.kde.org/kdev-python.git/commit/?id=836545d195e14c0df7e753726f7b5faecaa3e454

https://cgit.kde.org/kdev-python.git/commit/?id=d42f13080fbadf3d9afbe5516bed4fe4507f6203

Couldn't reproduce a crash even after reparsing tens of thousands of files, so I'm arbitrarily blaming it on a cache problem (declarations created before b553d75e might have weird ranges). Do complain if it breaks...

I did some more testing and it crashed...not sure why yet :-/

https://cgit.kde.org/kdev-python.git/commit/?id=1a7d3c139ab7b4ab439d757a95a5d7d83575f801