Homebrew-core: Qt 5.6.1 fails to build with '--with-examples'

[RandomDSdevel]

Please follow the general troubleshooting steps first:

• [x] Ran brew update and retried your prior step?
• [x] Ran brew doctor, fixed as many issues as possible and retried your prior step?
• [ ] If you're seeing permission errors tried running sudo chown -R $(whoami) $(brew --prefix)? (N/A)

My logs are here and here. My first attempt's logs are just those created by Homebrew itself upon failure, but my second attempt's logs also include the following files:

• reinstall.log
• doctor.log
• config.log

These files contain the results of redirecting standard output and standard error to them from the commands…:

• HOMEBREW_MAKE_JOBS=1 brew reinstall -vd --build-from-source qt5 --with-dbus --with-docs --with-examples --with-mysql
• brew doctor -vd
• brew config

…, respectively.

[cjrh]

Same error during make (as seen in the "first attempt" make log above):

make[2]: *** [sub-widgets-make_first] Error 2
make[1]: *** [sub-examples-make_first] Error 2
make: *** [module-qtbase-make_first] Error 2

[UniqMartin]

@RandomDSdevel Both failures can be attributed to the same cause, it's just that due to parallel build the error is a bit more difficult to spot in the first log file. In both cases the build should be fine if you remove --with-examples from your invocation.

Apparently a regression has been introduced in 5.6.1 that causes the build of the examples to fail, which is with 99% certainty an upstream bug. I'd encourage you to check with upstream whether this is a known problem and whether there's already a fix and if not, file a bug report.

@cjrh Quite possible this has the same cause and the same workaround, but with this little information that's really just a wild guess. Only if you can provide the information as requested in our Troubleshooting guide I can try to provide a more satisfying answer.

[cjrh]

@UniqMartin I reckon your analysis is correct regarding examples. This behaviour is certainly new and I agree it's most likely an upstream problem.

[RandomDSdevel]

@UniqMartin: I had a little time to do a few searches of Qt's bug tracker but wasn't able to find any upstream mentions of this bug. Either nobody's filed it yet or I just didn't find it. Either I'll file it later or you or @cjrh can do that first if either of you get to it before I do.

[RandomDSdevel]

@UniqMartin: Now that I've had time to do a somewhat more extensive search of [Qt's bug tracker] to see if anybody else had already filed the upstream bug corresponding to this issue, it turns out that my and @cjrh's problem was fixed in commit 58408ffa1b9c0b42a1719d3c8a4d4c62dec4fce6 in response to QTBUG-54110. Would modifying the qt5 formula so that it applies this patch be something easy that I could do myself, or do Homebrew maintainers like yourself usually manage Qt formula updates?

[UniqMartin]

@RandomDSdevel Thanks for finding that! If you feel comfortable preparing a PR for that, go ahead (and let me know). The fix is still needed for Qt 5.7.0 (currently being prepared in #2087), so we would need to apply it anyway. Otherwise I wouldn't mind extracting that patch and fixing it directly myself as this isn't part of the default configuration and thus doesn't need to go through the build bot and all that stuff.

[RandomDSdevel]

@UniqMartin: You're welcome! Yes, I do feel comfortable preparing and submitting a PR for this as soon as I can since the changes I'll need to make look simple enough for even a Homebrew novice like myself to contribute, so I'll get around to doing that as soon as I can, then.

[RandomDSdevel]

@UniqMartin: As you can see by the notifications that GitHub produced above when I referenced this issue from my pull request and the commit it contains, they're now both ready! I'll need some help refining my contribution, though, since the patch doesn't get applied properly (perhaps we're missing some prerequisite commits from upstream here?)

[UniqMartin]

Build with --with-examples should now succeed after updating the formula via the usual brew update. Thanks to @RandomDSdevel for contributing a PR after finding the relevant bug report and matching fix!

[RandomDSdevel]

@UniqMartin: Once again, you're very welcome!