add py3.8 patches from pyside 5.14

[looooo]

Checklist

• Used a fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[jschueller]

Nice, I tried with 5.12.x but the patches did not apply.
Which commits did you backport ? did it apply cleanly ? If the apply cleanly its best to retain the git messages. Else it would be good to mention which commits the patches are from.

[looooo]

I used these two commits:

commit 095c6437152aca85174662cdb78cfd583c03f5e4
Author: Christian Tismer <tismer@stackless.com>
Date:   Thu Dec 5 11:08:50 2019 +0100

    Optimize the Python 3.8 refcount fix a tiny bit
    
    This change uses the fact that our workaround to temporarily remove
    the Py_TPFLAGS_METHOD_DESCRIPTOR flag uses the "mro" function
    of PyType_Type, which never will change.
    Therefore, the static keyword makes sure that this function lookup
    happens only once.
    
    Change-Id: I44b74556da1fac2596c81339af30cb66218276e2
    Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>

commit 40a1a6bbeea82d8ccaaf7ec840ae8e0b8a183bc6
Author: Christian Tismer <tismer@stackless.com>
Date:   Tue Nov 26 11:54:37 2019 +0100

    Fix Python 3.8 problems
    
    This patch fixes some refcounting problems with Python 3.8 .
    One incompatible change was announced in the what's new
    document, but actually there were two more problems which
    were not explicitly mentioned but took much time to sort out.
    
    The patch is compatible with the limited API changes
    (tested with debug build and API error disabled).
    It is also independent of the Python version which is
    full Limited API support.
    
    For more info, see the documentation mentioned below.
    
    The flag error is circumvented now! We either find a better
    solution or leave it as it is. For now this is ok.
    
    Fixes: PYSIDE-939
    Change-Id: Iff4a9816857a6ebe86efd4b654d8921e4e464939
    Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
    Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>

And added these changes to make the compilation work again:

diff --git a/sources/shiboken2/libshiboken/basewrapper.cpp b/sources/shiboken2/libshiboken/basewrapper.cpp
index 1e3b7406..5824dd66 100644
--- a/sources/shiboken2/libshiboken/basewrapper.cpp
+++ b/sources/shiboken2/libshiboken/basewrapper.cpp
@@ -506,7 +506,8 @@ PyObject *SbkObjectTypeTpNew(PyTypeObject *metatype, PyObject *args, PyObject *k
     // PYSIDE-939: This is a temporary patch that circumvents the problem
     // with Py_TPFLAGS_METHOD_DESCRIPTOR until this is finally solved.
     PyObject *ob_PyType_Type = reinterpret_cast<PyObject *>(&PyType_Type);
-    static PyObject *mro = PyObject_GetAttr(ob_PyType_Type, Shiboken::PyName::mro());
+    static PyObject *const mro_string = Shiboken::String::fromCString("mro");
+    static PyObject *mro = PyObject_GetAttr(ob_PyType_Type, mro_string);
     auto hold = Py_TYPE(mro)->tp_flags;
     Py_TYPE(mro)->tp_flags &= ~Py_TPFLAGS_METHOD_DESCRIPTOR;
     auto *newType = reinterpret_cast<SbkObjectType *>(type_new(metatype, args, kwds));

[hmaarrfk]

you are still patching up 5.13.1, do you want to patch up 5.12.6?

[looooo]

I haven't tried to do so. I used the last available build (pyside2 5.13.1) as it was working nicely for me.

[looooo]

@hmaarrfk we used pyside2 5.13 with qt5.12 for a while without too much trouble. So I thought this combination is fine. Are there any known issues of pyside2 5.13 + qt5.12?

[hmaarrfk]

its just not so supported, see discussion in #61

Not sure what version your patches were targetting.

[looooo]

@hmaarrfk While pyside2 5.13 might not be supported to work with qt 5.12 (officially) I didn't see any bigger issues with this combination. I also saw this combination used in other distros, so I think we should stick to pyside5.13 for now and switch back once issues are reported.

[hmaarrfk]

I'm not sure where the static keyword is comming from. It seems to me that it isn't in the 5.13.1 or 5.13.2 source

[hmaarrfk]

OHHHHHHHHHHHHH. You need to flip 0003 and 0004.

[hmaarrfk]

you seem to have accidentally deleted the patches.

[hmaarrfk]

hmm, didn't know i had rights to your branch. lets see what happens.

[hmaarrfk]

Nice thanks for doing this. Should all versions of python be patched with this?

[jschueller]

Patches can be applied for all versions, no need for selectors.

[isuruf]

What does this do?

[jschueller]

this is a work-around as pyside messes up system includes and does not find xcb headers installed in the prefix

[jschueller]

Patches can be applied for all versions, no need for selectors.