resolving Qt4 slot conflicts: +egl +debug +aqua

ExecutorElassus

So, background: I am trying to install the git version of frescobaldi, which doesn't (but should!) have a "live" ebuild in portage. This version uses qt5, and depends on python-poppler-qt5 for music display (both frescobaldi and python-poppler-qt5 can be found here).

I'm encountering a number of problems, and I'm hoping one of you can help.

1) compilation of python-poppler-qt5 fails with the following:

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

When did you last run --depclean?

Switching to GCC-5 on a system that bad in shape is a very bad idea.

ExecutorElassus · Posted: Mon Jan 16, 2017 2:25 pm Post subject:

last depclean run was about a year ago, maybe more.

Should I go through and clear out old packages first?

Cheers,

EE

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

Yes. Make sure emerge --depclean -p is clean (and does not remove packages you want), check that eclean -d distfiles does not report any unavailable installed versions, check that emerge -uvpDN world does not come up with new dependencies or conflicts. Then make the GCC switch and rebuild. Basically that's what you do before any scary update.

ExecutorElassus · Posted: Mon Jan 16, 2017 3:17 pm Post subject:

alright, I'll proceed. But what exactly do you mean by "make the GCC switch"? gcc-config has been reporting the 5.* compiler as default for over a year now. What remains to be "switched"?

Cheers,

EE

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

ExecutorElassus · Posted: Mon Jan 16, 2017 6:51 pm Post subject:

I ran that command against libstdc++.so.5 back when the switch to gcc-5* first happened, but not since.

Anyway, now I've gone back and manually decleaned a bunch of packages (depclean, for some reason, misses a lot of dependencies that equery is able to find). However, running revdep-rebuild still results in the slot conflict with qtgui, namely:

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

ct85711 · Veteran Joined: 27 Sep 2005 Posts: 1791

well first off, generally you should not add the debug USE flag. Beyond that, looking at qtgui ebuild, it does indeed depend on those packages listed above on all of them having the same USE flags. I generally like using the -v to help show all the use flags for the packages.

ExecutorElassus · Posted: Tue Jan 17, 2017 7:11 am Post subject:

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

ExecutorElassus · Posted: Tue Jan 17, 2017 10:40 am Post subject:

I was just using nano, which has its own search function.

As for python-poppler-qt5: can you explain a bit more what you're suggesting I do with src_prepare? I've never used flag-o-matic, and I don't see a file called "src_prepare" in python-poppler-qt5's directories.

Cheers,

EE

mir3x · Guru Joined: 02 Jun 2012 Posts: 455

>> 1) compilation of python-poppler-qt5 fails with the following:

Qt5.7 requires c++11, gcc6.x automatically uses it.
gcc5.x and 4.x not.
All u need is to pass -std=c++11 to CXXFLAGS
CXXFLAGS=-std=c++11 emerge python-poppler-qt5

( seems its not ebuild so u can try hacking setup.py and add CXXFLAGS there or something like that )

>> 3) just remove all qt4 confilicting packages, they will be pulled back if needed:D ( dont add aqua or debug)
_________________
Sent from Windows

asturm · Developer Joined: 05 Apr 2007 Posts: 8936

ExecutorElassus · Posted: Tue Jan 17, 2017 1:59 pm Post subject:

Ah, I see the source of the confusion. I'm not using an ebuild for python-poppler-qt5 (so far as I know, nobody's made one [but they should!]). I am cloning directly from the git repository (here), and then running the setup instructions given in the INSTALL file. This is, then, entirely a python3-run installation, so I'm not sure how to instruct the installer to use gcc with C++11 enabled. Is there a way to do that?

Cheers,

EE

Hu · Moderator Joined: 06 Mar 2007 Posts: 21624

I suggest that you be the one to make such an ebuild, since you are trying to use a package that would clearly benefit from one. Wrapping it in an ebuild would ensure that each time you start over, you get a clean baseline, free of any interference from failed prior runs. The installer should have selected C++11 mode automatically. Since it does not, and since it claims to require C++11 mode, I would classify that as a bug in upstream's installer. As others in the post said, you could probably fix this by adding -std=c++11 or -std=gnu++11 to the compiler flags. If upstream follows convention, that would be $CXXFLAGS.

ExecutorElassus · Posted: Wed Jan 18, 2017 9:37 am Post subject:

well, ok, though I have no experience with writing ebuilds at all. So I just cribbed from a couple others, and here's what I have so far:

This is /usr/local/portage/dev-python/python-poppler-qt5/python-poppler-qt5-9999.ebuild

ExecutorElassus · Posted: Wed Jan 18, 2017 9:50 am Post subject:

UPDATE: whoops, needed to add 'git-r3' to the inherit line. Now I get a little further. Now I get this:

Hu · Moderator Joined: 06 Mar 2007 Posts: 21624

Both, I think. You misspelled a command in the ebuild: you wrote disutils, but the proper spelling is distutils. However, I do not think that mistake would cause the upstream build to fail in this way. You probably need help from someone familiar with this package.

ExecutorElassus · Posted: Fri Jan 27, 2017 2:36 pm Post subject:

Thank you for the tip. I fixed my ebuild.

However, now it still fails, and so I think it's an issue with upstream. Unfortunately, "upstream" is actually just one guy who also has a day job, so the issue I filed on the github page has thus far gone unanswered.

Would it help if I submitted the ebuild to gentoo's bugzilla, and see if other devs can have a crack at it?

Cheers,

EE

Hu · Moderator Joined: 06 Mar 2007 Posts: 21624

I think it is fine to ask for help. I am not sure if that is the right way to seek help for this part of Gentoo.

jok · n00b Joined: 26 Jan 2008 Posts: 22 Location: Finland

ExecutorElassus · Posted: Fri Feb 17, 2017 11:02 pm Post subject:

holy crap, that fixed it!

thanks for the help. So now, I have an ebuild that (so far as I can tell) successfully installs python-poppler-qt5, which is something that apparently the Frescobaldi package requires. Would this be something useful to post somewhere?

Cheers,

EE

jok · n00b Joined: 26 Jan 2008 Posts: 22 Location: Finland

Since you already have a github account, I'd suggest making a pull request (or two (or three)) to get the sip fix, python-poppler-qt5 and frescobaldi-3 in the main tree.
I know I'd be happy to have one less Qt4 anchor installed :wink:

Hu · Moderator Joined: 06 Mar 2007 Posts: 21624

If you have the time to get it merged, I recommend doing so. Eventually, someone else will want to use this, and it would help them if your fix is already present by default when they arrive.