I do, just let me know what's the preferred merge workflow: push to 4.1 then merge to master or push to master and cherry-pick in 4.1?

We usually do master, then cherry-pick to 4.1

Eek, I tested this on Kubuntu, where it worked fine. On OpenSuse and on the nightly builds we get:

CMake Error at plugins/extensions/pykrita/sip/CMakeLists.txt:15 (if):

if given arguments:

  "050a01" "STRGREATER_EQUAL" "050b00"

Unknown arguments specified

• Configuring incomplete, errors occurred!

Oh, STRGREATER_EQUAL requires cmake 3.7. If can be safely changed to STRGREATER, since PYQt 5.11.0 was withdrawn upstream so no distro shound be using it.

Argh... I've changed the version check into VERSION_GREATER. and now it builds find on my local opensuse system, but the appimage builds are still broken:

https://binary-factory.kde.org/job/Krita_Nightly_Appimage_Build/157/consoleText
https://binary-factory.kde.org/job/Krita_Stable_Appimage_Build/57/consoleText

We are comparing string-formatted versions, so you should use STRGREATER, not VERSION_GREATER. Otherwise this will give unexpected results: on my system the if() evalulates to false even though I have PyQt 5.11, and in the appimage it seems to evaluate to TRUE with PyQt 5.9.

STRGREATER is available in cmake 2.8.12 which is Krita's required version, so it should be fine.

8ee0958ae632e49773b79f837daaf40305777d06 is still wrong: ${PYQT5_VERSION} is the alphanumeric version string, it can't be compared with 5.11.0. Either

if(${PYQT5_VERSION_STR} VERSION_GREATER 5.11.0)

or

if(${PYQT5_VERSION} STRGREATER "050b00")

should work (see cmake/modules/FindPyQt5.cmake)

Let's try if(${PYQT5_VERSION_STR} VERSION_GREATER 5.11.0) for now.

Hm.... I've now moved to a laptop with the latest Neon and Qt 5.11, and now I get a build error:

[ 97%] Generating krita/sipkritapart0.cpp

sip: Unable to find file "QtCore/QtCoremod.sip"
plugins/extensions/pykrita/sip/CMakeFiles/python_module_PyKrita_krita.dir/build.make:93: recipe for target 'plugins/extensions/pykrita/sip/krita/sipkritapart0.cpp' failed
make[2]: * [plugins/extensions/pykrita/sip/krita/sipkritapart0.cpp] Error 1
make[2]: * Deleting file 'plugins/extensions/pykrita/sip/krita/sipkritapart0.cpp'
CMakeFiles/Makefile2:29863: recipe for target 'plugins/extensions/pykrita/sip/CMakeFiles/python_module_PyKrita_krita.dir/all' failed
make[1]: * [plugins/extensions/pykrita/sip/CMakeFiles/python_module_PyKrita_krita.dir/all] Error 2
make[1]: * Waiting for unfinished jobs....

No idea about this error (the location of the QtCore/QtCoremod.sip hasn't changed in 5.11 so it should be found as before, and in any case it should be unrelated to this patch).

However, I noticed that Debian has patched PyQt to revert upstream's decision of using a private sip:

https://salsa.debian.org/python-team/modules/pyqt5/commit/ecc66ca38179d0dbcb65212421dec11bc1e252e7

so Krita plugins loading will fail for them as it did for Arch before this patch. I personally don't understand upstream's decision and I think it makes things unnecessarily harder for packagers, but if every distro starts doing their own thing it will be a mess for downstream projects like Krita to support every possible scenario.

Argh... I guess the only thing I can do now is add a manual override cmake flag.

In D14061#293606, @rempt wrote:

Argh... I guess the only thing I can do now is add a manual override cmake flag.

There may be a way around it: it looks like PyQt5.Qt.PYQT_CONFIGURATION["sip_flags"] contains the sipname that was used to compile PyQt, so Krita should use that sipname for its bindings. I will try to write a patch.

okay, that would be great.

New patch at https://phabricator.kde.org/D14182