| Age | Commit message (Collapse) | Author |
|---|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Upon closer look the fix is subpar as did not formerly notice the
removed bit had a replacement looking only at the header (without
the commit). Not that this is an issue with what it's used with at
the moment (afaik only qutebrowser really uses PyQt6's sql) so will
stick with it until next release.
Kind of wonder how Arch got it to build without patches given there
is a "Qt 6.6 rebuild" PyQt6 commit, maybe it actually used 6.5 given
6.6 is still in experimental (not that this would be an issue given
it works at runtime with 6.6 then).
Retried with final version (formerly 6.6.0-rc) without patch, and it
still fails.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Both fail to build. No revbump needed given it should be usable
with 6.6 if it was built against an older version like 6.5.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Believe symbol issues had in the past were more of a fluke than
the norm, not to mention this does not actually use _p.h headers.
Takes a while to rebuild and forcing a rebuild on every Qt bump
may not be worthwhile (back when added this, Qt used 0/a.b rather
than 0/a.b.c so it was not that bad). May revert if get issues again.
(would still cause one more rebuild but -r1 happened in an earlier
commit either way)
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
qtbase[widgets] forces it on qtdeclarative either way, but could
"potentially" lead portage to believe it can schedule enabling
widgets on qtdeclarative after PyQt6.
No revbump needed for this edge case.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
QtPdf is enabled in qtwebengine-6.5.3+.
Not using IUSE=pdf given qtwebengine tend to need special treatment,
and this makes it easier to mask the USE globally on arches where it
is not supported. Furthermore IUSE=pdf is default on some profiles and
would pull the heavy qtwebengine for users that may not want it.
Revbumping at same time as 6.5.3 given binding operator will cause
a rebuild either way.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Nothing that needs this in PyQt6 currently, but there is no harm
in adding off-by-default support now that these are packaged.
Only ones still missing are:
- QtLocation: still disabled in project.py, so waiting on upstream
- QtPdf: needs enabling & testing in qtwebengine first
- QtRemoteObjects: thought to package this but failing to find anything
that is going to need it currently, will wait until someone requests
(remoteobjects sound interesting but seemingly scarcely used)
- QtAxContainer: this is for Windows (ActiveX), so nothing to do here
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Which I am part of, so nothing changes beside
being consistent with other PyQt* packages.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Not a todo, had a closer look at the feature and it's
enabled for "APPLE OR ANDROID OR WASM", so not really
our thing.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
No tests provided like other packages from this chain from this
upstream, so using qutebrowser[qt6]'s tests as reference to know
if it works with py3.12.
qutebrowser did have 3.12 issues with 6.5.1, which got fixed in
6.5.2 -- so from this assuming that the chain's latest versions
have likely been tested by upstream with 3.12 now and should be
mostly safe to add.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Turns out that this *can* use clang++ but only if qtbase was built with
clang++ as well, otherwise it still uses g++ even with CXX=clang++
But given the former, there is little reason to enforce the dependency.
Ideally would use CXX but that would require more elaborate testing to
know what it is and what qtbase was built with (perhaps if ever add the
hack to qmake-utils.eclass).
No revbump given PyQt* bindings are a slowish build and depcleaning
gcc is a bit niche, the few interested can handle this manually until
next bump.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Seems to rarely be an issue but hit symbol problems when using
PyQt6-6.4.2 with Qt6.5 without a rebuild (otherwise works with
6.5 just fine).
Could happen with 6.6+ too, albeit mitigated by PyQt6 bumps happening
around same time and forcing the rebuild either way.
Going to skip revbump for now given the above and being a slow build
(plus it's masked), uncertain whether it'll even be needed for 6.6.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Didn't really make a difference before, but now it does,
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Interestly this seems to only use libglvnd directly if
gles2-only (for libGLESv2), otherwise uses GL seemingly
only through qtbase.
Not worth a revbump, still masked and libglvnd is still
guaranteed through qtbase either way -- just a nitpick
from iwdevtools reporting the DT_NEEDED entry.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
On top of disabling the QtOpenGL bindings with USE=-opengl,
also need to fully disable the OpenGL feature given it's
also used in QtGui.
Still not sure if this is all handled right for gles2-only,
the logic is misleading.
Closes: https://bugs.gentoo.org/904676
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Switch to using /usr/bin/qmake6 which is now provided by qtbase.
Also bump QT_PV to use $(ver_cut 1-2), not always necessary but
this makes it unnecessary to keep track of new Qt features usage.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: David Seifert <soap@gentoo.org>
|
|
Forgot IUSE=opengl is not default on qtbase, will leave that
one to desktop profiles to enable it.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Feel this is the generally sane way to handle defaults
as don't get far without base modules.
This adds opengl and printsupport on top of the rest, both
already used by qutebrowser (albeit still misses qml which
needs a non-base module).
Exception for testlib as it's afaik useless outside tests.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
qt-6.4.0 is only in the ::qt overlay the moment, but this works
just fine with 6.3.2 that I can see. Adding but may be better to
wait to match 6.x as a precaution in the future, notably confuses
1 qutebrowser test but it has no real consequence.
Eitherway, given it's masked not too woried about this and want
to get some testing out of these things to know effects of mismatch.
PyQt6-6.4.0 adds support for QtTextToSpeech and QtPdf but neither
can be enabled at the moment.
- pdf: from qtwebengine:6 but is currently disabled in its ebuild
- speech: not available in <qt-6.4.0 and qtspeech:6 is not packaged
in the qt overlay either yet, skips the need to bump QT_PV to
6.4:6 here for now
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
qmake6 will be either moved or symlinked in /usr/bin (so could
`type -P qmake6`), but the former would break this until updated
unless use no-suffix qmake (same file).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Still unsure what want to do with this.
Right now using qtbase's defaults as a baseline rather than
what qutebrowser needs (aka pulls no extra deps, albeit each
USE still add a lot to compile time), but if the qtbase
package is ever split this may be questionable.
PyQt5 sets nothing which is simple, but without anything set
these packages are near useless and ensures users will need
to set something.
May revisit when this gain other consumers and/or when getting
close to unmasking.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
Was meant to match to skip MY_P (like dev-python often does), but
on second thought it doesn't feel entirely right here given
(ignoring pypi tarballs) official website doesn't use underscore
and old PyQt5-sip didn't either.
Could probably skip pkgmove (qt6 still masked), but feels messy
not to so doing it anyway.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
May still be rough around the edges, currently only tested with
qutebrowser so please report if missing anything for other consumers.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
