C/C++: LGTM.com run failed on PR but CI succeeds after changes to CMakeLists.txt

[VolkerEnderlein]

After changes to a CMakeLists.txt an LGTM.com run on the PR failed, but all other GitHub CI runs succeeded.

The execute_process commands are needed to query qmake for QT_INSTALL_PREFIX and QT_INSTALL_PLUGINS but they do not populate the CMake variables. The build log at line 577 shows that.

See here for additional information (build logs etc.).

[VolkerEnderlein]

It seems the behaviour is caused by the qmake-wrapper. When called with -query QT_INSTALL_PREFIX argument it should return the path on stdout but it returns it with additional text on stderr:

[2021-11-04 09:36:41] [build-stderr] CMake Error at src/plugins/CMakeLists.txt:55 (file):
[2021-11-04 09:36:41] [build-stdout] -- QT_QMAKE_EXECUTABLE: "/usr/lib/x86_64-linux-gnu/qt5/bin/qmake"
[2021-11-04 09:36:41] [build-stderr]   file RELATIVE_PATH must be passed a full path to the directory:
[2021-11-04 09:36:41] [build-stdout] -- QT_INSTALL_PREFIX: ""
[2021-11-04 09:36:41] [build-stdout] -- QT_INSTALL_PREFIX_RESULT: "0"
[2021-11-04 09:36:41] [build-stdout] -- QT_INSTALL_PREFIX_ERROR: "
[2021-11-04 09:36:41] [build-stdout] qmake-wrapper: first run
[2021-11-04 09:36:41] [build-stdout] /usr
[2021-11-04 09:36:41] [build-stdout] "

[adityasharad]

Thanks for the report and the investigation you've carried out already. From a brief look it does indeed look like our qmake-wrapper is redirecting stdout from qmake to its own stderr instead of stdout. I've mentioned this to our CodeQL C/C++ team, will report back next week once we have an idea of what we can change here.

[podsvirov]

Three week later...

Hello to all! Any news there?

[VolkerEnderlein]

@adityasharad Any news on the issue?

[adityasharad]

The qmake-wrapper used in the container images on LGTM.com has been updated to fix this problem. Thanks for your patience. Please reopen this and let us know if you continue to see the problem!

[podsvirov]

Good news, will check at new case!

[VolkerEnderlein]

Thanks @adityasharad we'll try the updated version. ... After trying out: I can confirm it is now working as expected. Thanks again for the support. Keep up the good work!