x11/py-qt5-opengl: Enable QOpenGLFunctions, do not install stub files, and strip shared libraries

[jrm]

Summary

The OpenGLFunctions are required by qutebrowser.

Test Plan

portlint -C: no new warnings/errors
poudriere testport 10/11 i386/amd64: passes with no qa warnings

[jrm]

Another cosmetic change would be to put CONFIGURE_ARGS= and PYQT_DIST= after the USES and USE_x block to conform to the ordering specified in Chapter 14. of the Porter's Handbook.

[tcberner]

x11/py-qt5-opengl/Makefile

34

why not just ${STRIP_CMD$ ${STAGEDIR}${PREFIX}/lib/${PYTHON_VERSION}/site-packages/PyQt5/*.so

[tcberner]

In D12982#269633, @jrm wrote:

Another cosmetic change would be to put CONFIGURE_ARGS= and PYQT_DIST= after the USES and USE_x block to conform to the ordering specified in Chapter 14. of the Porter's Handbook.

I think that can be done later, as all the PyQt ports probably can use that cleanup -- but I'm not mad, if you already include it here :)

[sunpoet]

x11/py-qt5-opengl/Makefile

38

Why not use PYTHON_SITELIBDIR?

[jrm]

Uses *.so and PYTHON_SITELIBDIR when stripping shared libraries

[jrm]

I'm not clear why we are not just specifying USE_PYQT=sip_build, then allowing consumers to specify the complete required list for USE_PYQT.

[tcberner]

In D12982#269701, @jrm wrote:

I'm not clear why we are not just specifying USE_PYQT=sip_build, then allowing consumers to specify the complete required list for USE_PYQT.

What do you mean?

[jrm]

I mean why doesn't this port simply use USE_PYQT=sip_build and then consumers of py-qt5 modules, like www/qutebrowser, specify the full list of modules it requires?

[tcberner]

In D12982#269711, @jrm wrote:

I mean why doesn't this port simply use USE_PYQT=sip_build and then consumers of py-qt5 modules, like www/qutebrowser, specify the full list of modules it requires?

I'd say because you can install py-qt5 standalone without any applications using it (say for in house stuff) -- which should then also be able to run, and therefore pull in the required "real" Qt5 foo.

[jrm]

In D12982#269739, @tcberner wrote:

I'd say because you can install py-qt5 standalone without any applications using it (say for in house stuff) -- which should then also be able to run, and therefore pull in the required "real" Qt5 foo.

That makes sense, but why would a user install py-qt5-opengl to get many (but not all) py-qt5/qt5 modules. For users that want a more complete installation of py-qt5, maybe we could create a meta-port that pulls in the modules.

[tcberner]

There already exists such a meta port :) -- Maybe I don't quite understand what the problem is here :)

[mat]

x11/py-qt5-opengl/Makefile

9–11

Out of order. See Chapter 14. Order of Variables in Port Makefiles.

16–17

missing a tab after =.

[jrm]

x11/py-qt5-opengl/Makefile

9–11

See the comments on this diff where I suggested this, but was asked to wait.

[jrm]

Add missing TAB and align values in block

[jrm]

In D12982#269929, @tcberner wrote:

Maybe I don't quite understand what the problem is here :)

[mat]

x11/py-qt5-opengl/Makefile

9–11

Well, "we'll do it later" usually means it'll never be done, so unless someone has a todo list with this item on it with an actual deadline that is not three years in the future, it is probably better to do this now.

[tcberner]

x11/py-qt5-opengl/Makefile

9–11

I noticed some other "stuff to cleanup" for the PyQt ports, so it's on a Todo list here :)

[tcberner]

That being said, and thought about, @mat is right. There is no good reason to wait for me to find time to do them all in one go, so please add the cleanups already :)

[jrm]

• Order variable blocks according to Chapter 14. of the Porter's Handbook
• Add a comment about LICENSE to prevent portlint warnings and for clarity for others viewing the Makefile

[tcberner]

x11/py-qt5-opengl/Makefile

10

which reminds me to commit this... give me some minutes :D