Re: UPDATE: Qt 5.13.2
On 2020/03/08 16:45, Landry Breuil wrote:
> On Sun, Mar 08, 2020 at 12:59:30PM +, Stuart Henderson wrote:
> > On 2020/03/08 12:40, Stuart Henderson wrote:
> > > py-sip-qt5 attached for completeness, it is the same as the last one from
> > > Landry. OK sthen@ to import that unhooked, then we can hook it to the
> > > build
> > > when switching over.
> >
> > py-sip-qt5 needs this added:
> >
> > MAKE_FLAGS += CC="${CC}" CXX="${CXX}"
> >
>
> i had it fixed locally by amending the existing patch copied from
> py-sip:
I prefer MAKE_FLAGS because it actually uses what is set through ports,
though IIRC this is all very hit-and-miss with qt ports anyway.
> but i'm fine with MAKE_FLAGS too - fixed version attached.
>
> still uncertain on the naming ? x11/py-sip-qt5 ? upstream names it
> PyQt5-sip...
> py-qt5-sip ? who cares ?
I don't :) I am ok with your current name.
> my bulk is at 3550 so far, the only thing to fix is the version
> dependency in devel/spyder/spyder which has
> x11/py-qt5${MODPY_FLAVOR}<5.12 for some reason:
>
> Error: @depend x11/py-qt5,python3:py3-qt5-<5.12:py3-qt5-5.13.2
> pattern py3-qt5-<5.12 doesn't match default py3-qt5-5.13.2
So for spyder.. upstream for the version in ports has <=5.12, which was later
amended to <5.13 (49c6c2aa86 "Setup.py: Fix PyQt5 pinning conditions").
The port Makefile has <5.12 which doesn't quite match either of these.
Upstream's current position:
https://github.com/spyder-ide/spyder/issues/9829#issuecomment-574749134
Can you try removing the version dependency and see if spyder works anyway?
Looking at the github issues you mentioned it seems the most likely thing to
break is theme icons for the "Spyder 3" theme (tools/prefs/scroll down).
> the only other failures i have are fetching
> devel/libtalloc:talloc-2.1.16.tar.gz databases/tdb:tdb-1.3.18.tar.gz but
> those are totally unrelated, https://download.samba.org/pub/talloc/
> returns a 403 to dpb, but i can fetch it locally so i guess it's just a
> fluke.
old ftp(1) :)
> definitely agree we should go ahead will all this :)
Re: UPDATE: Qt 5.13.2
On Sat, Mar 07 2020, Jeremie Courreges-Anglas wrote: [...] > - some -examples subpackages probably need a PLIST refresh. Note that > qttools,-main packaged fine, I'm restarting a build to see whether > -examples also packaged properly Packages fine, no PLIST change except for an additional "@so " in pkg/PLIST-webview. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
Re: UPDATE: Qt 5.13.2
On Sun, Mar 08, 2020 at 12:40:11PM +, Stuart Henderson wrote:
> I'm doing a test build now. (Have disabled a couple of large ports that
> are nothing to do with qt5, but otherwise a standard bulk).
>
> Here is the complete diff that I'm building with, it includes WANTLIB
> fixes in dependent ports as spotted by cwen@. I have commented-out
> qtlottie from x11/qt5/Makefile because I don't have the files and it
> isn't needed by anything, so it can wait until the main part is done
> (I don't want anything else adding delays :)
>
> py-sip-qt5 attached for completeness, it is the same as the last one from
> Landry. OK sthen@ to import that unhooked, then we can hook it to the build
> when switching over.
>
> If my build doesn't find any more problems then I think we should get
> this lot committed ASAP.
my bulk is at 3550 so far, the only thing to fix is the version
dependency in devel/spyder/spyder which has
x11/py-qt5${MODPY_FLAVOR}<5.12 for some reason:
Error: @depend x11/py-qt5,python3:py3-qt5-<5.12:py3-qt5-5.13.2
pattern py3-qt5-<5.12 doesn't match default py3-qt5-5.13.2
the only other failures i have are fetching
devel/libtalloc:talloc-2.1.16.tar.gz databases/tdb:tdb-1.3.18.tar.gz but
those are totally unrelated, https://download.samba.org/pub/talloc/
returns a 403 to dpb, but i can fetch it locally so i guess it's just a
fluke.
definitely agree we should go ahead will all this :)
Re: UPDATE: Qt 5.13.2
On Sun, Mar 08, 2020 at 12:59:30PM +, Stuart Henderson wrote:
> On 2020/03/08 12:40, Stuart Henderson wrote:
> > py-sip-qt5 attached for completeness, it is the same as the last one from
> > Landry. OK sthen@ to import that unhooked, then we can hook it to the build
> > when switching over.
>
> py-sip-qt5 needs this added:
>
> MAKE_FLAGS += CC="${CC}" CXX="${CXX}"
>
i had it fixed locally by amending the existing patch copied from
py-sip:
$OpenBSD: patch-specs_openbsd-g++,v 1.2 2017/03/02 15:10:15 espie Exp $
Index: specs/openbsd-g++
--- specs/openbsd-g++.orig
+++ specs/openbsd-g++
@@ -7,7 +7,7 @@ TEMPLATE= app
CONFIG += qt warn_on release link_prl
QT += core gui
-QMAKE_CC = gcc
+QMAKE_CC = cc
QMAKE_LEX = flex
QMAKE_LEXFLAGS =
QMAKE_YACC = yacc
@@ -22,7 +22,7 @@ QMAKE_CFLAGS_SHLIB= -fPIC
QMAKE_CFLAGS_YACC = -Wno-unused -Wno-parentheses
QMAKE_CFLAGS_THREAD= -pthread
-QMAKE_CXX = g++
+QMAKE_CXX = c++
QMAKE_CXXFLAGS = $$QMAKE_CFLAGS
QMAKE_CXXFLAGS_DEPS= $$QMAKE_CFLAGS_DEPS
QMAKE_CXXFLAGS_WARN_ON = $$QMAKE_CFLAGS_WARN_ON
but i'm fine with MAKE_FLAGS too - fixed version attached.
still uncertain on the naming ? x11/py-sip-qt5 ? upstream names it PyQt5-sip...
py-qt5-sip ? who cares ?
going to start an amd64 bulk with your patchset, but i agree we should commit
this soonish to avoid more churn. I'm globally okay with the patchset anyway,
and all the consumers i tested were fine (except tortoisehg but i suppose its
local as it works for rafael)
Landry
py-sip-qt5_2.tgz
Description: application/tar-gz
Re: UPDATE: Qt 5.13.2
On 2020/03/08 12:40, Stuart Henderson wrote:
> py-sip-qt5 attached for completeness, it is the same as the last one from
> Landry. OK sthen@ to import that unhooked, then we can hook it to the build
> when switching over.
py-sip-qt5 needs this added:
MAKE_FLAGS += CC="${CC}" CXX="${CXX}"
Re: UPDATE: Qt 5.13.2
On Sat Mar 07, 2020 at 03:31:04PM +0100, Jeremie Courreges-Anglas wrote:
> On Thu, Feb 27 2020, Rafael Sadowski wrote:
> > Hi All
> >
> > After truly hard work, I am excited to write this email. The
> > following diff contains the 3 ports which are all dependent on each
> > other.
> >
> > - devel/py-sip
> > - x11/py-qt5
> > - x11/qt5
> >
> > Almost all consumers fixed. For some you will find patches on
> > ports@, for some, private maintainer emails are out. math/octave is not
> > happy now but maybe an update helps. I'll take care of this monster.
> >
> > What's the news?
> >
> > - Most of the work was done in qtbase.
> > - The qtbase port comes with vulkan and zstd support by default enabled.
> > I hope this will works fine with !{amd64,i386} arches.
> > - s/c++11/c++17/
> > - Qtdoc (docs) is broken again, I think I am not that bad again and can
> > be fixed later.
> > - Bump all shred lib and cleanup SHARED_LIBS.
> > - Many cleaning jobs in the Makefiles.
> > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at this.
> > - patch-qmake_generators_unix_unixmake_cpp
> > -- That was the biggest problem, at the p2k19 I decided to solve by:
> > "Transform /usr/ports/pobj/xxx/lib/libQt5Core.so into
> > -L/usr/ports/pobj/xxx/build-amd64/lib -lQt5Core" ... works!
> > - ... and much more...
> >
> > What do *we* need next? I tested on amd64 i386, so we need build-tests
> > on powerpc, sparc64, aarch64. KEEP IN MIND to delete Qt5.9 first before
> > you build/update 5.13. (pkg_delete qtbase does the job). This behaviour
> > was already broken in the old version and I did not find time to fix it.
> >
> > I wouldn't have thought at the beginning of the year that we would
> > probably see Qt 5.13 in OpenBSD 6.7. Thanks for all the support and
> > especially for the motivation not to give up.
> >
> > Looking forward for feedback.
>
> On sparc64 I tried to build, using the first diff in this thread:
Merci jca@!
> - x11/qt5/qt!(canvas3d|webengine|webview)/
> - devel/py-sip
> - x11/py-qt5
> - security/keepassxc
> - graphics/nomacs
> - devel/qt-creator
> - mail/trojita
> - cad/qcad
> - print/lyx
> - print/TeXmacs
> - www/minitube
> - meta/qt5
> - devel/qbs
> - x11/lxqt
> - x11/cool-retro-term
> - editors/tiled
>
> (avoiding kf5, items selected with no scientific method)
>
> Plus some py-qt5 consumers:
> - devel/tortoisehg
Tested with Landry's patch set, works fine here.
> - devel/git-cola
> - net/onionshare
> - editors/py-qscintilla
>
> dpb term.log, slightly edited:
> --8<--
> Hosts: localhost
> I=769 B=3 Q=0 T=2 F=0 !=10
> L=qt/qtcanvas3d-everywhere-src-5.13.2.tar.xz.dist
> E=devel/tortoisehg
> editors/py-qscintilla
> mail/trojita
> cad/qcad!
> x11/qt5/qttools,,-examples!
> x11/qt5/qtmultimedia,,-examples!
> x11/qt5/qtdeclarative,,-examples!
> x11/qt5/qtcanvas3d:qt/qtcanvas3d-everywhere-src-5.13.2.tar.xz
> x11/qt5/qtcanvas3d
> x11/qt5/qtcanvas3d,-examples
> -->8--
>
> Logs for all mentioned ports can be found at cvs:~jca/tmp/
>
> Here's a digest:
>
> - (known) editors/py-qscintilla and devel/tortoisehg fail because of
> py-sip etc, I did not follow all the mails between you, Landry and
> Stuart.
Fixed by Landry, tested today.
>
> - meta/qt5 still depends on x11/qt5/qtcanvas3d (-main and -examples)
> even though qtcanvas3d has been unhooked
ACK; Will be commit with x11/qt5
>
> - some -examples subpackages probably need a PLIST refresh. Note that
> qttools,-main packaged fine, I'm restarting a build to see whether
> -examples also packaged properly
>
ACK; I'll take a look.
> - qcad fails because the ports infrastructure can't find cups:
> "Missing library for cups>=0.0" and indeed x11/qt5/base doesn't depend
> on print/cups,-libs any more. Either qcad needs to be adjusted or
> something is wrong here.
ACK; Let's add print/cups,-libs to LIB_DEPENDS, OK?
>
> - trojita failure (with surrounding warnings elided):
> --8<--
> /usr/ports/pobj/trojita-0.7/trojita-0.7/tests/Imap/test_Imap_BodyParts.cpp:56:7:
> error: redefinition of 'char* QTest::toString(const T&) [with T =
> QModelIndex]'
> char *toString(const QModelIndex )
>^~~~
> In file included from /usr/local/include/X11/qt5/QtTest/QtTest:11,
> from
> /usr/ports/pobj/trojita-0.7/trojita-0.7/tests/Imap/test_Imap_BodyParts.cpp:24:
> /usr/local/include/X11/qt5/QtTest/qtest.h:132:25: note: 'char*
> QTest::toString(const T&) [with T = QModelIndex]' previously declared here
> template<> inline char *toString(const QModelIndex )
> ^~~~
Already fixed in r1.32.
> -->8--
>
> I've been sitting on this report for a few days, sorry, hope this helps
> anyway.
>
> --
> jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
>
Re: UPDATE: Qt 5.13.2
On Thu, Feb 27 2020, Rafael Sadowski wrote:
> Hi All
>
> After truly hard work, I am excited to write this email. The
> following diff contains the 3 ports which are all dependent on each
> other.
>
> - devel/py-sip
> - x11/py-qt5
> - x11/qt5
>
> Almost all consumers fixed. For some you will find patches on
> ports@, for some, private maintainer emails are out. math/octave is not
> happy now but maybe an update helps. I'll take care of this monster.
>
> What's the news?
>
> - Most of the work was done in qtbase.
> - The qtbase port comes with vulkan and zstd support by default enabled.
> I hope this will works fine with !{amd64,i386} arches.
> - s/c++11/c++17/
> - Qtdoc (docs) is broken again, I think I am not that bad again and can
> be fixed later.
> - Bump all shred lib and cleanup SHARED_LIBS.
> - Many cleaning jobs in the Makefiles.
> - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at this.
> - patch-qmake_generators_unix_unixmake_cpp
> -- That was the biggest problem, at the p2k19 I decided to solve by:
> "Transform /usr/ports/pobj/xxx/lib/libQt5Core.so into
> -L/usr/ports/pobj/xxx/build-amd64/lib -lQt5Core" ... works!
> - ... and much more...
>
> What do *we* need next? I tested on amd64 i386, so we need build-tests
> on powerpc, sparc64, aarch64. KEEP IN MIND to delete Qt5.9 first before
> you build/update 5.13. (pkg_delete qtbase does the job). This behaviour
> was already broken in the old version and I did not find time to fix it.
>
> I wouldn't have thought at the beginning of the year that we would
> probably see Qt 5.13 in OpenBSD 6.7. Thanks for all the support and
> especially for the motivation not to give up.
>
> Looking forward for feedback.
On sparc64 I tried to build, using the first diff in this thread:
- x11/qt5/qt!(canvas3d|webengine|webview)/
- devel/py-sip
- x11/py-qt5
- security/keepassxc
- graphics/nomacs
- devel/qt-creator
- mail/trojita
- cad/qcad
- print/lyx
- print/TeXmacs
- www/minitube
- meta/qt5
- devel/qbs
- x11/lxqt
- x11/cool-retro-term
- editors/tiled
(avoiding kf5, items selected with no scientific method)
Plus some py-qt5 consumers:
- devel/tortoisehg
- devel/git-cola
- net/onionshare
- editors/py-qscintilla
dpb term.log, slightly edited:
--8<--
Hosts: localhost
I=769 B=3 Q=0 T=2 F=0 !=10
L=qt/qtcanvas3d-everywhere-src-5.13.2.tar.xz.dist
E=devel/tortoisehg
editors/py-qscintilla
mail/trojita
cad/qcad!
x11/qt5/qttools,,-examples!
x11/qt5/qtmultimedia,,-examples!
x11/qt5/qtdeclarative,,-examples!
x11/qt5/qtcanvas3d:qt/qtcanvas3d-everywhere-src-5.13.2.tar.xz
x11/qt5/qtcanvas3d
x11/qt5/qtcanvas3d,-examples
-->8--
Logs for all mentioned ports can be found at cvs:~jca/tmp/
Here's a digest:
- (known) editors/py-qscintilla and devel/tortoisehg fail because of
py-sip etc, I did not follow all the mails between you, Landry and
Stuart.
- meta/qt5 still depends on x11/qt5/qtcanvas3d (-main and -examples)
even though qtcanvas3d has been unhooked
- some -examples subpackages probably need a PLIST refresh. Note that
qttools,-main packaged fine, I'm restarting a build to see whether
-examples also packaged properly
- qcad fails because the ports infrastructure can't find cups:
"Missing library for cups>=0.0" and indeed x11/qt5/base doesn't depend
on print/cups,-libs any more. Either qcad needs to be adjusted or
something is wrong here.
- trojita failure (with surrounding warnings elided):
--8<--
/usr/ports/pobj/trojita-0.7/trojita-0.7/tests/Imap/test_Imap_BodyParts.cpp:56:7:
error: redefinition of 'char* QTest::toString(const T&) [with T = QModelIndex]'
char *toString(const QModelIndex )
^~~~
In file included from /usr/local/include/X11/qt5/QtTest/QtTest:11,
from
/usr/ports/pobj/trojita-0.7/trojita-0.7/tests/Imap/test_Imap_BodyParts.cpp:24:
/usr/local/include/X11/qt5/QtTest/qtest.h:132:25: note: 'char*
QTest::toString(const T&) [with T = QModelIndex]' previously declared here
template<> inline char *toString(const QModelIndex )
^~~~
-->8--
I've been sitting on this report for a few days, sorry, hope this helps
anyway.
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
Re: UPDATE: Qt 5.13.2
On 2020/03/06 08:49, Landry Breuil wrote: > > grepping, the py-qt5 ports that reference webengine are: > > what kind of grepping, source code or requirements in setup.py ? source
Re: UPDATE: Qt 5.13.2
On Fri, Mar 06, 2020 at 08:49:38AM +0100, Landry Breuil wrote:
> On Thu, Mar 05, 2020 at 10:12:41PM +, Stuart Henderson wrote:
> > On 2020/03/05 17:49, Landry Breuil wrote:
> > > On Thu, Mar 05, 2020 at 05:10:51PM +0100, Landry Breuil wrote:
> > > > On Thu, Mar 05, 2020 at 02:14:42PM +, Stuart Henderson wrote:
> > > > > Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK
> > > > > sthen@
> > > > > to import that.
> > > >
> > > > in the end its completely fucked up, because tortoisehg (py2 only) fails
> > > > to build with this py-qt5 without its sip counterpart, so i duplicated
> > > > the py-sip port as py-sip-qt5 (can of course change the name) with a
> > > > 4.19.19 version, as the 12.7.1 on pypi explicitely targets py3 only.
> > > > argh. dont look at the port itself, i know there are CVS dirs,
> > > > CONFIGURE_ARGS is what matters with --no-tools --sip-module.
> > > >
> > > > while here, removed some churn from pyqt5 PLIST diff (thePYQT_VERSION
> > > > var serves no purpose), and send the right diffs.
> > > >
> > > > oh boy that's a pain.
> > >
> > > but it works, been able to build tortoisehg with it, and now building
> > > qgis.
> > >
> > > on a sidenote we can see that QtWebEngine disappeared from py-qt5 port,
> > > its now in
> > > https://www.riverbankcomputing.com/software/pyqtwebengine/download
> > >
> >
> > grepping, the py-qt5 ports that reference webengine are:
>
> what kind of grepping, source code or requirements in setup.py ?
>
> > git-cola
> > spyder
> > qgis
> > qutebrowser (marked BROKEN)
> >
> > ... and spyder has a specific version dependency:
> >
> > x11/py-qt5${MODPY_FLAVOR}<5.12
>
> i guess this is related to
> https://github.com/spyder-ide/spyder/issues/8748 and
> https://github.com/spyder-ide/spyder/issues/8747, an update will
> probably loosen this requirement..
>
> > sigh, does this thing ever end? :)
>
> well, that seems to be the case for every qt upgrade, a complete
> ecosystem that needs to be updated altogether. But i think we're slowly
> getting there.. now i'm ccing the ppl who use/maintain those pyqt5 apps,
> at that point it has to be a collective effort for testing. I guess i'll
> do a quick port for pyqtwebengine, unless rafael already has one ?
after actually looking at it, i think we'll be fine without a
pyqtwebengine port, and anyway we dont have qtwebengine. So far, my
understanding is that py-qt5 5.9 only ships some .sip files for
qtwebengine bindings and no actual lib, so the potential consumers might
not actually use it (and might only need to be told "dont look for
it"..).
git-cola bundles a copy of qtpy which is where the qtwebengine bits are
looked for (from my understanding) .. and spyder imports
QtWebEngineWidgets from qtpy too. I really think we're fine.
(if ppl want to do some runtime testing to save the time i'm putting pkgs on
https://packages.rhaalovely.net/wip/amd64/)
completely unrelated, but i wonder if we could move the spyder port to
py3 only to reduce the whack-a-mole needed for testing.
Landry
Re: UPDATE: Qt 5.13.2
On Thu, Mar 05, 2020 at 10:12:41PM +, Stuart Henderson wrote:
> On 2020/03/05 17:49, Landry Breuil wrote:
> > On Thu, Mar 05, 2020 at 05:10:51PM +0100, Landry Breuil wrote:
> > > On Thu, Mar 05, 2020 at 02:14:42PM +, Stuart Henderson wrote:
> > > > Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK
> > > > sthen@
> > > > to import that.
> > >
> > > in the end its completely fucked up, because tortoisehg (py2 only) fails
> > > to build with this py-qt5 without its sip counterpart, so i duplicated
> > > the py-sip port as py-sip-qt5 (can of course change the name) with a
> > > 4.19.19 version, as the 12.7.1 on pypi explicitely targets py3 only.
> > > argh. dont look at the port itself, i know there are CVS dirs,
> > > CONFIGURE_ARGS is what matters with --no-tools --sip-module.
> > >
> > > while here, removed some churn from pyqt5 PLIST diff (thePYQT_VERSION
> > > var serves no purpose), and send the right diffs.
> > >
> > > oh boy that's a pain.
> >
> > but it works, been able to build tortoisehg with it, and now building
> > qgis.
> >
> > on a sidenote we can see that QtWebEngine disappeared from py-qt5 port,
> > its now in
> > https://www.riverbankcomputing.com/software/pyqtwebengine/download
> >
>
> grepping, the py-qt5 ports that reference webengine are:
what kind of grepping, source code or requirements in setup.py ?
> git-cola
> spyder
> qgis
> qutebrowser (marked BROKEN)
>
> ... and spyder has a specific version dependency:
>
> x11/py-qt5${MODPY_FLAVOR}<5.12
i guess this is related to
https://github.com/spyder-ide/spyder/issues/8748 and
https://github.com/spyder-ide/spyder/issues/8747, an update will
probably loosen this requirement..
> sigh, does this thing ever end? :)
well, that seems to be the case for every qt upgrade, a complete
ecosystem that needs to be updated altogether. But i think we're slowly
getting there.. now i'm ccing the ppl who use/maintain those pyqt5 apps,
at that point it has to be a collective effort for testing. I guess i'll
do a quick port for pyqtwebengine, unless rafael already has one ?
Landry
Re: UPDATE: Qt 5.13.2
On 2020/03/05 17:49, Landry Breuil wrote:
> On Thu, Mar 05, 2020 at 05:10:51PM +0100, Landry Breuil wrote:
> > On Thu, Mar 05, 2020 at 02:14:42PM +, Stuart Henderson wrote:
> > > Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK sthen@
> > > to import that.
> >
> > in the end its completely fucked up, because tortoisehg (py2 only) fails
> > to build with this py-qt5 without its sip counterpart, so i duplicated
> > the py-sip port as py-sip-qt5 (can of course change the name) with a
> > 4.19.19 version, as the 12.7.1 on pypi explicitely targets py3 only.
> > argh. dont look at the port itself, i know there are CVS dirs,
> > CONFIGURE_ARGS is what matters with --no-tools --sip-module.
> >
> > while here, removed some churn from pyqt5 PLIST diff (thePYQT_VERSION
> > var serves no purpose), and send the right diffs.
> >
> > oh boy that's a pain.
>
> but it works, been able to build tortoisehg with it, and now building
> qgis.
>
> on a sidenote we can see that QtWebEngine disappeared from py-qt5 port,
> its now in
> https://www.riverbankcomputing.com/software/pyqtwebengine/download
>
grepping, the py-qt5 ports that reference webengine are:
git-cola
spyder
qgis
qutebrowser (marked BROKEN)
... and spyder has a specific version dependency:
x11/py-qt5${MODPY_FLAVOR}<5.12
sigh, does this thing ever end? :)
Re: UPDATE: Qt 5.13.2
On Thu, Mar 05, 2020 at 05:10:51PM +0100, Landry Breuil wrote: > On Thu, Mar 05, 2020 at 02:14:42PM +, Stuart Henderson wrote: > > Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK sthen@ > > to import that. > > in the end its completely fucked up, because tortoisehg (py2 only) fails > to build with this py-qt5 without its sip counterpart, so i duplicated > the py-sip port as py-sip-qt5 (can of course change the name) with a > 4.19.19 version, as the 12.7.1 on pypi explicitely targets py3 only. > argh. dont look at the port itself, i know there are CVS dirs, > CONFIGURE_ARGS is what matters with --no-tools --sip-module. > > while here, removed some churn from pyqt5 PLIST diff (thePYQT_VERSION > var serves no purpose), and send the right diffs. > > oh boy that's a pain. but it works, been able to build tortoisehg with it, and now building qgis. on a sidenote we can see that QtWebEngine disappeared from py-qt5 port, its now in https://www.riverbankcomputing.com/software/pyqtwebengine/download
Re: UPDATE: Qt 5.13.2
On Thu, Mar 05, 2020 at 02:14:42PM +, Stuart Henderson wrote:
> Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK sthen@
> to import that.
in the end its completely fucked up, because tortoisehg (py2 only) fails
to build with this py-qt5 without its sip counterpart, so i duplicated
the py-sip port as py-sip-qt5 (can of course change the name) with a
4.19.19 version, as the 12.7.1 on pypi explicitely targets py3 only.
argh. dont look at the port itself, i know there are CVS dirs,
CONFIGURE_ARGS is what matters with --no-tools --sip-module.
while here, removed some churn from pyqt5 PLIST diff (thePYQT_VERSION
var serves no purpose), and send the right diffs.
oh boy that's a pain.
Landry
py-sip-qt5.tgz
Description: application/tar-gz
Index: Makefile
===
RCS file: /cvs/ports/devel/py-sip/Makefile,v
retrieving revision 1.49
diff -u -r1.49 Makefile
--- Makefile19 Feb 2020 15:27:01 - 1.49
+++ Makefile5 Mar 2020 12:49:24 -
@@ -5,7 +5,7 @@
# XXX py-sip updates FREQUENTLY break other ports
# XXX BE ESPECIALLY CAREFUL WHEN TESTING UPDATES!
-MODPY_EGG_VERSION =4.19.18
+MODPY_EGG_VERSION =4.19.19
EPOCH= 0
DISTNAME= sip-${MODPY_EGG_VERSION}
PKGNAME= py-${DISTNAME}
Index: distinfo
===
RCS file: /cvs/ports/devel/py-sip/distinfo,v
retrieving revision 1.20
diff -u -r1.20 distinfo
--- distinfo19 Feb 2020 15:27:01 - 1.20
+++ distinfo5 Mar 2020 12:49:24 -
@@ -1,2 +1,2 @@
-SHA256 (sip-4.19.18.tar.gz) = wL2GOADtmxXcrUd8QBfNtz+oBcJZCLAkBWSt101pfh4=
-SIZE (sip-4.19.18.tar.gz) = 1056840
+SHA256 (sip-4.19.19.tar.gz) = VDa2Gnj0jH6AeOk6a1lFOtM3gPgMZE5fOvOflL4e3kQ=
+SIZE (sip-4.19.19.tar.gz) = 1058422
Index: Makefile
===
RCS file: /cvs/ports/x11/py-qt5/Makefile,v
retrieving revision 1.21
diff -u -r1.21 Makefile
--- Makefile19 Feb 2020 15:27:46 - 1.21
+++ Makefile5 Mar 2020 16:02:52 -
@@ -2,8 +2,8 @@
COMMENT= python Qt5 bindings
-V= 5.9.2
-DISTNAME= PyQt5_gpl-${V}
+V= 5.13.2
+DISTNAME= PyQt5-${V}
PKGNAME= py-qt5-${V}
CATEGORIES=x11 devel
@@ -14,31 +14,35 @@
# GPLv3 with exceptions for use with certain named open-source licenses
PERMIT_PACKAGE=Yes
-WANTLIB = ${COMPILER_LIBCXX} ${MODPY_WANTLIB}
-WANTLIB += dbus-1 glib-2.0 intl m pthread pulse pulse-mainloop-glib
-WANTLIB += GL Qt5Bluetooth Qt5Core Qt5DBus Qt5Designer
-WANTLIB += Qt5Gui Qt5Help Qt5Multimedia Qt5MultimediaWidgets Qt5Network
-WANTLIB += Qt5OpenGL Qt5Positioning Qt5PrintSupport Qt5Qml Qt5Quick
-WANTLIB += Qt5QuickWidgets Qt5Sensors Qt5SerialPort Qt5Sql Qt5Svg
-WANTLIB += Qt5Test Qt5WebChannel Qt5WebKit Qt5WebKitWidgets Qt5WebSockets
-WANTLIB += Qt5Widgets Qt5X11Extras Qt5Xml Qt5XmlPatterns
-WANTLIB += Qt5Location Qt5Nfc
+WANTLIB += ${COMPILER_LIBCXX} GL Qt5Bluetooth Qt5Core Qt5DBus
+WANTLIB += Qt5Designer Qt5Gui Qt5Help Qt5Location Qt5Multimedia
+WANTLIB += Qt5MultimediaWidgets Qt5Network Qt5NetworkAuth Qt5Nfc
+WANTLIB += Qt5OpenGL Qt5Positioning Qt5PositioningQuick Qt5PrintSupport
+WANTLIB += Qt5Qml Qt5Quick Qt5QuickWidgets Qt5RemoteObjects Qt5Sensors
+WANTLIB += Qt5SerialPort Qt5Sql Qt5Svg Qt5Test Qt5WebChannel Qt5WebKit
+WANTLIB += Qt5WebKitWidgets Qt5WebSockets Qt5Widgets Qt5X11Extras
+WANTLIB += Qt5Xml Qt5XmlPatterns dbus-1 glib-2.0 intl m pulse
+WANTLIB += pulse-mainloop-glib ${MODPY_WANTLIB}
FLAVORS = python3
FLAVOR ?=
BUILD_DEPENDS= ${RUN_DEPENDS}
-RUN_DEPENDS= devel/py-sip${MODPY_FLAVOR}>=4.19.1v0 \
+RUN_DEPENDS= devel/py-sip${MODPY_FLAVOR}>=4.19.19v0 \
+ x11/py-sip-qt5${MODPY_FLAVOR}>=4.19.19 \
x11/dbus-python${MODPY_FLAVOR}
+
LIB_DEPENDS= ${MODPY_LIB_DEPENDS} \
x11/qt5/qtconnectivity \
x11/qt5/qtdeclarative \
x11/qt5/qtlocation \
+ x11/qt5/qtnetworkauth \
x11/qt5/qtmultimedia \
x11/qt5/qtserialport \
x11/qt5/qtsensors \
x11/qt5/qtsvg \
x11/qt5/qttools \
+ x11/qt5/qtremoteobjects \
x11/qt5/qtwebchannel \
x11/qt5/qtwebkit \
x11/qt5/qtwebsockets \
@@ -47,11 +51,16 @@
NO_TEST= Yes
-MASTER_SITES= ${MASTER_SITE_SOURCEFORGE:=pyqt/}
+MASTER_SITES= https://www.riverbankcomputing.com/static/Downloads/PyQt5/${V}/
MODULES= lang/python \
x11/qt5
+.if !${FLAVOR:Mpython3}
+BUILD_DEPENDS += devel/py-enum34
+RUN_DEPENDS += devel/py-enum34
+.endif
+
CONFIGURE_STYLE=simple
CONFIGURE_SCRIPT=${MODPY_BIN} configure.py
CONFIGURE_ARGS=--verbose \
@@ -63,6 +72,7 @@
--vendorid-incdir="${MODPY_INCDIR}" \
--vendorid-libdir="${MODPY_LIBDIR}" \
Re: UPDATE: Qt 5.13.2
Looking better! Stray blank line in py-qt5-sip DESCR otherwise OK sthen@
to import that.
(I'd slightly prefer to have a port directory that matches the python
module name (so py-PyQt5-sip or py-PyQt5_sip) but OTOH it does match
our nonstandard naming py-qt5, so...meh :)
> - make it a dependency of py3-qt5 only, a bit less churn in the PLIST
> but keep the _v{$MODPY_MAJOR_VERSION} bit as with py2 it ends up with _v2
ah right, thanks :)
> with this patchset ive been able to build py-qt5 & py-qscintilla, needs
> runtime testing with tortoisehg, and im still building py3-qt5 &
> py3-qscintilla to finally testbuild qgis.
>
> more eyes/feedback welcome :)
I have been testing some of the non-py-qt5 ports using Qt at runtime and
not had problems with them yet.
My py-qt5 builds are still running so no further tests on that yet.
Re: UPDATE: Qt 5.13.2
On Thu, Mar 05, 2020 at 09:05:39AM +0100, Landry Breuil wrote:
> On Thu, Mar 05, 2020 at 08:00:52AM +0100, Landry Breuil wrote:
> > On Tue, Mar 03, 2020 at 10:12:55PM +, Stuart Henderson wrote:
> > > On 2020/03/03 18:16, Rafael Sadowski wrote:
> > > > On Mon Mar 02, 2020 at 06:09:01PM +0100, Rafael Sadowski wrote:
> > > > > On Mon Mar 02, 2020 at 03:41:55PM +0100, Landry Breuil wrote:
> > > > > > On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote:
> > > > > > > On 2020/02/29 18:38, Landry Breuil wrote:
> > > > > > > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson
> > > > > > > > wrote:
> > > > > > > > > On 2020/02/29 07:27, Landry Breuil wrote:
> > > > > > > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski
> > > > > > > > > > wrote:
> > >
> > > This looks wrong. py-qt5 is not the only thing using py-sip, I think you
> > > need either a separate port or a flavour for this.
> >
> > i had a hard time wrapping my head around this, and i agree with stuart
> > here. adding --sip-module somewhat renames the python sip module to
> > PyQt5_sip, which might not be what sip consumers expect, ie at least the
> > ones via x11/py-qt4 which depends on the python2 py-sip. Building py-qt4
> > here with the diff, but it would rather require runtime testing with
> > consmers, or import tests via a python cli, etc. only building is not
> > enough.
> >
> > an option would be to use --sip-module only in the python3 flavor
> > for py-sip, since py-qt5 is the only consumer. but that would require a
> > bit more SUBST work in the py-sip port. will eventually try.
>
> Here's a py-sip diff that does that, ie change nothing for the
> unflavoured version, and use what seems needed for py3/qt5. built both
> flavors here, will build py-qt5 and py-qt4 and some consumers. feedback
> & testing welcome :)
and after more discussion with sthen, a different approach:
- stop messing up with py-sip, just update it
- port py-qt5-sip separately, its py3 only as it requires py>=3.5
- make it a dependency of py3-qt5 only, a bit less churn in the PLIST
but keep the _v{$MODPY_MAJOR_VERSION} bit as with py2 it ends up with _v2
with this patchset ive been able to build py-qt5 & py-qscintilla, needs
runtime testing with tortoisehg, and im still building py3-qt5 &
py3-qscintilla to finally testbuild qgis.
more eyes/feedback welcome :)
Landry
Index: Makefile
===
RCS file: /cvs/ports/devel/py-sip/Makefile,v
retrieving revision 1.42
diff -u -r1.42 Makefile
--- Makefile28 Apr 2019 20:51:37 - 1.42
+++ Makefile26 Jun 2019 16:05:19 -
@@ -5,9 +5,9 @@
# XXX Be particularly careful when testing updates.
# API breaks often.
-DISTNAME= sip-4.19.3
+MODPY_EGG_VERSION =4.19.17
+DISTNAME= sip-${MODPY_EGG_VERSION}
PKGNAME= py-${DISTNAME}
-REVISION = 4
CATEGORIES=devel
HOMEPAGE= http://www.riverbankcomputing.co.uk/software/sip/intro
@@ -19,7 +19,7 @@
COMPILER = base-clang ports-gcc base-gcc
-MASTER_SITES= ${MASTER_SITE_SOURCEFORGE:=pyqt/}
+MASTER_SITES=
https://www.riverbankcomputing.com/static/Downloads/sip/${MODPY_EGG_VERSION}/
MODULES= lang/python
@@ -43,6 +43,12 @@
SUBST_VARS += MODPY_LIB_SUFFIX
pre-configure:
${SUBST_CMD} ${WRKSRC}/configure.py
+
+# mk_distinfo.py tries to open sip_bin but it hasnt been renamed yet
+.if ${FLAVOR:Mpython3}
+pre-fake:
+ cd ${PREFIX}/bin && ln -sf sip sip${MODPY_BIN_SUFFIX}
+.endif
post-install:
${MODPY_BIN} ${MODPY_LIBDIR}/compileall.py ${WRKINST}${MODPY_SITEPKG}
Index: distinfo
===
RCS file: /cvs/ports/devel/py-sip/distinfo,v
retrieving revision 1.16
diff -u -r1.16 distinfo
--- distinfo17 Jul 2017 08:34:39 - 1.16
+++ distinfo26 Jun 2019 16:05:19 -
@@ -1,2 +1,2 @@
-SHA256 (sip-4.19.3.tar.gz) = dA34RPgMxF3MmyMpSpJJKSO8QDzojmjDV4PyfBd8S3Q=
-SIZE (sip-4.19.3.tar.gz) = 1007481
+SHA256 (sip-4.19.17.tar.gz) = ErzY9NX+78EFvAddEsUJDueD9zgHKFY8kbi5XQ7EXfM=
+SIZE (sip-4.19.17.tar.gz) = 1056156
Index: pkg/PLIST
===
RCS file: /cvs/ports/devel/py-sip/pkg/PLIST,v
retrieving revision 1.9
diff -u -r1.9 PLIST
--- pkg/PLIST 17 Jul 2017 08:34:39 - 1.9
+++ pkg/PLIST 26 Jun 2019 16:05:19 -
@@ -1,10 +1,14 @@
@comment $OpenBSD: PLIST,v 1.9 2017/07/17 08:34:39 zhuk Exp $
@bin bin/sip${MODPY_BIN_SUFFIX}
include/python${MODPY_VERSION}${MODPY_LIB_SUFFIX}/sip.h
-lib/python${MODPY_VERSION}/site-packages/sip.pyi
${MODPY_COMMENT}lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}/
lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipconfig.${MODPY_PYC_MAGIC_TAG}pyc
lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipdistutils.${MODPY_PYC_MAGIC_TAG}pyc
+lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/
Re: UPDATE: Qt 5.13.2
On Thu, Mar 05, 2020 at 08:00:52AM +0100, Landry Breuil wrote:
> On Tue, Mar 03, 2020 at 10:12:55PM +, Stuart Henderson wrote:
> > On 2020/03/03 18:16, Rafael Sadowski wrote:
> > > On Mon Mar 02, 2020 at 06:09:01PM +0100, Rafael Sadowski wrote:
> > > > On Mon Mar 02, 2020 at 03:41:55PM +0100, Landry Breuil wrote:
> > > > > On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote:
> > > > > > On 2020/02/29 18:38, Landry Breuil wrote:
> > > > > > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote:
> > > > > > > > On 2020/02/29 07:27, Landry Breuil wrote:
> > > > > > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski
> > > > > > > > > wrote:
> >
> > This looks wrong. py-qt5 is not the only thing using py-sip, I think you
> > need either a separate port or a flavour for this.
>
> i had a hard time wrapping my head around this, and i agree with stuart
> here. adding --sip-module somewhat renames the python sip module to
> PyQt5_sip, which might not be what sip consumers expect, ie at least the
> ones via x11/py-qt4 which depends on the python2 py-sip. Building py-qt4
> here with the diff, but it would rather require runtime testing with
> consmers, or import tests via a python cli, etc. only building is not
> enough.
>
> an option would be to use --sip-module only in the python3 flavor
> for py-sip, since py-qt5 is the only consumer. but that would require a
> bit more SUBST work in the py-sip port. will eventually try.
Here's a py-sip diff that does that, ie change nothing for the
unflavoured version, and use what seems needed for py3/qt5. built both
flavors here, will build py-qt5 and py-qt4 and some consumers. feedback
& testing welcome :)
my py-qt5 diff still stands, but might need the fixes stuart suggested
about _v3 at least.
Landry
Index: Makefile
===
RCS file: /cvs/ports/devel/py-sip/Makefile,v
retrieving revision 1.49
diff -u -r1.49 Makefile
--- Makefile19 Feb 2020 15:27:01 - 1.49
+++ Makefile5 Mar 2020 07:59:45 -
@@ -5,7 +5,7 @@
# XXX py-sip updates FREQUENTLY break other ports
# XXX BE ESPECIALLY CAREFUL WHEN TESTING UPDATES!
-MODPY_EGG_VERSION =4.19.18
+MODPY_EGG_VERSION =4.19.19
EPOCH= 0
DISTNAME= sip-${MODPY_EGG_VERSION}
PKGNAME= py-${DISTNAME}
@@ -41,14 +41,20 @@
NO_TEST= Yes
-SUBST_VARS += MODPY_LIB_SUFFIX
+SUBST_VARS += MODPY_LIB_SUFFIX SIP_MODULE SIP_SUBDIR
pre-configure:
${SUBST_CMD} ${WRKSRC}/configure.py
# mk_distinfo.py tries to open sip_bin but it hasnt been renamed yet
.if ${FLAVOR:Mpython3}
+SIP_MODULE = PyQt5_sip
+SIP_SUBDIR = PyQt5
+CONFIGURE_ARGS += --sip-module ${SIP_MODULE:S/_/./}
pre-fake:
cd ${PREFIX}/bin && ln -sf sip sip${MODPY_BIN_SUFFIX}
+.else
+SIP_SUBDIR =
+SIP_MODULE = sip
.endif
post-install:
Index: distinfo
===
RCS file: /cvs/ports/devel/py-sip/distinfo,v
retrieving revision 1.20
diff -u -r1.20 distinfo
--- distinfo19 Feb 2020 15:27:01 - 1.20
+++ distinfo5 Mar 2020 07:59:45 -
@@ -1,2 +1,2 @@
-SHA256 (sip-4.19.18.tar.gz) = wL2GOADtmxXcrUd8QBfNtz+oBcJZCLAkBWSt101pfh4=
-SIZE (sip-4.19.18.tar.gz) = 1056840
+SHA256 (sip-4.19.19.tar.gz) = VDa2Gnj0jH6AeOk6a1lFOtM3gPgMZE5fOvOflL4e3kQ=
+SIZE (sip-4.19.19.tar.gz) = 1058422
Index: pkg/PLIST
===
RCS file: /cvs/ports/devel/py-sip/pkg/PLIST,v
retrieving revision 1.11
diff -u -r1.11 PLIST
--- pkg/PLIST 27 Jan 2020 10:31:00 - 1.11
+++ pkg/PLIST 5 Mar 2020 07:59:45 -
@@ -1,15 +1,16 @@
@comment $OpenBSD: PLIST,v 1.11 2020/01/27 10:31:00 rsadowski Exp $
@bin bin/sip${MODPY_BIN_SUFFIX}
include/python${MODPY_VERSION}${MODPY_LIB_SUFFIX}/sip.h
+${MODPY_COMMENT}lib/python${MODPY_VERSION}/site-packages/${SIP_SUBDIR}/
+lib/python${MODPY_VERSION}/site-packages/${SIP_SUBDIR}/sip.pyi
+@so lib/python${MODPY_VERSION}/site-packages/${SIP_SUBDIR}/sip.so
+lib/python${MODPY_VERSION}/site-packages/${SIP_MODULE}-${MODPY_EGG_VERSION}.dist-info/
+lib/python${MODPY_VERSION}/site-packages/${SIP_MODULE}-${MODPY_EGG_VERSION}.dist-info/INSTALLER
+lib/python${MODPY_VERSION}/site-packages/${SIP_MODULE}-${MODPY_EGG_VERSION}.dist-info/METADATA
+lib/python${MODPY_VERSION}/site-packages/${SIP_MODULE}-${MODPY_EGG_VERSION}.dist-info/RECORD
${MODPY_COMMENT}lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}/
lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipconfig.${MODPY_PYC_MAGIC_TAG}pyc
lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipdistutils.${MODPY_PYC_MAGIC_TAG}pyc
-lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/
-lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/INSTALLER
-lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/METADATA
Re: UPDATE: Qt 5.13.2
On Tue, Mar 03, 2020 at 10:12:55PM +, Stuart Henderson wrote:
> On 2020/03/03 18:16, Rafael Sadowski wrote:
> > On Mon Mar 02, 2020 at 06:09:01PM +0100, Rafael Sadowski wrote:
> > > On Mon Mar 02, 2020 at 03:41:55PM +0100, Landry Breuil wrote:
> > > > On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote:
> > > > > On 2020/02/29 18:38, Landry Breuil wrote:
> > > > > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote:
> > > > > > > On 2020/02/29 07:27, Landry Breuil wrote:
> > > > > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> > > > >
> > > > > Looking at
> > > > > https://www.riverbankcomputing.com/static/Docs/PyQt5/installation.html
> > > > >
> > > > > "It is not necessary to install the PyQt5.sip module before building
> > > > > PyQt5 but it must be installed before PyQt5 can be used.
> > >
> > > Fu** I missed this in my diff! BIG SORRY
> > >
> > > > >
> > > > > The module is built using setuptools and is available from the
> > > > > PyQt5-sip
> > > > > project at PyPI. It uses setuptools as its build system and can be
> > > > > installed by pip or you can also unpack the sdist and install it by
> > > > > running its setup.py script."
> > > > >
> > > > > So it looks like this should probably be a separate port ..
> > > >
> > > > ah, so its now a separate port, that should be a RUN_DEPENDS for py-qt5
> > > > ? rafael, does it ring a bell to you ?
> > >
> > > Yeah, this is necessary, sorry for missing this in my last diff. It's
> > > hard to painful to create such large diffs!
> > >
> >
> > New diff which includes the missing py-sip bits. Not yet adjusted,
> > landrys py-qt5 diff.
> > +@so lib/python${MODPY_VERSION}/site-packages/PyQt5/sip.so
> > +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/
> > +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/INSTALLER
> > +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/METADATA
> > +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/RECORD
> > ${MODPY_COMMENT}lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}/
> >
> > lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipconfig.${MODPY_PYC_MAGIC_TAG}pyc
> >
> > lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipdistutils.${MODPY_PYC_MAGIC_TAG}pyc
> > -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/
> > -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/INSTALLER
> > -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/METADATA
> > -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/RECORD
> > -lib/python${MODPY_VERSION}/site-packages/sip.pyi
> > -@so lib/python${MODPY_VERSION}/site-packages/sip.so
> > lib/python${MODPY_VERSION}/site-packages/sipconfig.py
> > lib/python${MODPY_VERSION}/site-packages/sipdistutils.py
> > share/doc/${MODPY_PY_PREFIX}sip/
>
> This looks wrong. py-qt5 is not the only thing using py-sip, I think you
> need either a separate port or a flavour for this.
i had a hard time wrapping my head around this, and i agree with stuart
here. adding --sip-module somewhat renames the python sip module to
PyQt5_sip, which might not be what sip consumers expect, ie at least the
ones via x11/py-qt4 which depends on the python2 py-sip. Building py-qt4
here with the diff, but it would rather require runtime testing with
consmers, or import tests via a python cli, etc. only building is not
enough.
an option would be to use --sip-module only in the python3 flavor
for py-sip, since py-qt5 is the only consumer. but that would require a
bit more SUBST work in the py-sip port. will eventually try.
> PFRAG.python3 is not needed, these entries were fine how they were
> in PLIST with the ${MODPY_COMMENT} prefix.
100% agree
> Question: does the *existing* py-qt5 work with the new x11/qt5 pieces?
Sadly, i really doubt so.. but only runtime testing would tell.. such a
huge runtime testing matrix.. with tightly coupled version
dependencies..
Landry
Re: UPDATE: Qt 5.13.2
On 2020/03/03 18:16, Rafael Sadowski wrote:
> On Mon Mar 02, 2020 at 06:09:01PM +0100, Rafael Sadowski wrote:
> > On Mon Mar 02, 2020 at 03:41:55PM +0100, Landry Breuil wrote:
> > > On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote:
> > > > On 2020/02/29 18:38, Landry Breuil wrote:
> > > > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote:
> > > > > > On 2020/02/29 07:27, Landry Breuil wrote:
> > > > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> > > >
> > > > Looking at
> > > > https://www.riverbankcomputing.com/static/Docs/PyQt5/installation.html
> > > >
> > > > "It is not necessary to install the PyQt5.sip module before building
> > > > PyQt5 but it must be installed before PyQt5 can be used.
> >
> > Fu** I missed this in my diff! BIG SORRY
> >
> > > >
> > > > The module is built using setuptools and is available from the PyQt5-sip
> > > > project at PyPI. It uses setuptools as its build system and can be
> > > > installed by pip or you can also unpack the sdist and install it by
> > > > running its setup.py script."
> > > >
> > > > So it looks like this should probably be a separate port ..
> > >
> > > ah, so its now a separate port, that should be a RUN_DEPENDS for py-qt5
> > > ? rafael, does it ring a bell to you ?
> >
> > Yeah, this is necessary, sorry for missing this in my last diff. It's
> > hard to painful to create such large diffs!
> >
>
> New diff which includes the missing py-sip bits. Not yet adjusted,
> landrys py-qt5 diff.
>
>
> Index: devel/py-sip/Makefile
> ===
> RCS file: /cvs/ports/devel/py-sip/Makefile,v
> retrieving revision 1.49
> diff -u -p -r1.49 Makefile
> --- devel/py-sip/Makefile 19 Feb 2020 15:27:01 - 1.49
> +++ devel/py-sip/Makefile 3 Mar 2020 17:07:58 -
> @@ -5,7 +5,7 @@ COMMENT= automatically generate Python b
> # XXX py-sip updates FREQUENTLY break other ports
> # XXX BE ESPECIALLY CAREFUL WHEN TESTING UPDATES!
>
> -MODPY_EGG_VERSION = 4.19.18
> +MODPY_EGG_VERSION = 4.19.19
> EPOCH= 0
> DISTNAME=sip-${MODPY_EGG_VERSION}
> PKGNAME= py-${DISTNAME}
> @@ -33,6 +33,7 @@ CONFIGURE_ARGS= -b ${PREFIX}/bin \
> -d ${MODPY_SITEPKG} \
> -e ${MODPY_INCDIR} \
> -v ${PREFIX}/share/sip \
> + --sip-module PyQt5.sip \
> INCDIR=${LOCALBASE}/include \
> LIBDIR=${LOCALBASE}/lib \
> CC=${CC} CXX=${CXX} \
> Index: devel/py-sip/distinfo
> ===
> RCS file: /cvs/ports/devel/py-sip/distinfo,v
> retrieving revision 1.20
> diff -u -p -r1.20 distinfo
> --- devel/py-sip/distinfo 19 Feb 2020 15:27:01 - 1.20
> +++ devel/py-sip/distinfo 3 Mar 2020 17:07:58 -
> @@ -1,2 +1,2 @@
> -SHA256 (sip-4.19.18.tar.gz) = wL2GOADtmxXcrUd8QBfNtz+oBcJZCLAkBWSt101pfh4=
> -SIZE (sip-4.19.18.tar.gz) = 1056840
> +SHA256 (sip-4.19.19.tar.gz) = VDa2Gnj0jH6AeOk6a1lFOtM3gPgMZE5fOvOflL4e3kQ=
> +SIZE (sip-4.19.19.tar.gz) = 1058422
> Index: devel/py-sip/pkg/PLIST
> ===
> RCS file: /cvs/ports/devel/py-sip/pkg/PLIST,v
> retrieving revision 1.11
> diff -u -p -r1.11 PLIST
> --- devel/py-sip/pkg/PLIST27 Jan 2020 10:31:00 - 1.11
> +++ devel/py-sip/pkg/PLIST3 Mar 2020 17:07:58 -
> @@ -1,15 +1,16 @@
> @comment $OpenBSD: PLIST,v 1.11 2020/01/27 10:31:00 rsadowski Exp $
> @bin bin/sip${MODPY_BIN_SUFFIX}
> include/python${MODPY_VERSION}${MODPY_LIB_SUFFIX}/sip.h
> +lib/python${MODPY_VERSION}/site-packages/PyQt5/
> +lib/python${MODPY_VERSION}/site-packages/PyQt5/sip.pyi
> +@so lib/python${MODPY_VERSION}/site-packages/PyQt5/sip.so
> +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/
> +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/INSTALLER
> +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/METADATA
> +lib/python${MODPY_VERSION}/site-packages/PyQt5_sip-${MODPY_EGG_VERSION}.dist-info/RECORD
> ${MODPY_COMMENT}lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}/
>
> lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipconfig.${MODPY_PYC_MAGIC_TAG}pyc
>
> lib/python${MODPY_VERSION}/site-packages/${MODPY_PYCACHE}sipdistutils.${MODPY_PYC_MAGIC_TAG}pyc
> -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/
> -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/INSTALLER
> -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/METADATA
> -lib/python${MODPY_VERSION}/site-packages/sip-${MODPY_EGG_VERSION}.dist-info/RECORD
> -lib/python${MODPY_VERSION}/site-packages/sip.pyi
> -@so lib/python${MODPY_VERSION}/site-packages/sip.so
> lib/python${MODPY_VERSION}/site-packages/sipconfig.py
>
Re: UPDATE: Qt 5.13.2
On Mon Mar 02, 2020 at 03:41:55PM +0100, Landry Breuil wrote: > On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote: > > On 2020/02/29 18:38, Landry Breuil wrote: > > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote: > > > > On 2020/02/29 07:27, Landry Breuil wrote: > > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote: > > > > Looking at > > https://www.riverbankcomputing.com/static/Docs/PyQt5/installation.html > > > > "It is not necessary to install the PyQt5.sip module before building > > PyQt5 but it must be installed before PyQt5 can be used. Fu** I missed this in my diff! BIG SORRY > > > > The module is built using setuptools and is available from the PyQt5-sip > > project at PyPI. It uses setuptools as its build system and can be > > installed by pip or you can also unpack the sdist and install it by > > running its setup.py script." > > > > So it looks like this should probably be a separate port .. > > ah, so its now a separate port, that should be a RUN_DEPENDS for py-qt5 > ? rafael, does it ring a bell to you ? Yeah, this is necessary, sorry for missing this in my last diff. It's hard to painful to create such large diffs!
Re: UPDATE: Qt 5.13.2
On Mon, Mar 02, 2020 at 02:34:05PM +, Stuart Henderson wrote: > On 2020/02/29 18:38, Landry Breuil wrote: > > On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote: > > > On 2020/02/29 07:27, Landry Breuil wrote: > > > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote: > > Looking at > https://www.riverbankcomputing.com/static/Docs/PyQt5/installation.html > > "It is not necessary to install the PyQt5.sip module before building > PyQt5 but it must be installed before PyQt5 can be used. > > The module is built using setuptools and is available from the PyQt5-sip > project at PyPI. It uses setuptools as its build system and can be > installed by pip or you can also unpack the sdist and install it by > running its setup.py script." > > So it looks like this should probably be a separate port .. ah, so its now a separate port, that should be a RUN_DEPENDS for py-qt5 ? rafael, does it ring a bell to you ?
Re: UPDATE: Qt 5.13.2
On 2020/02/29 18:38, Landry Breuil wrote:
> On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote:
> > On 2020/02/29 07:27, Landry Breuil wrote:
> > > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> > > > On Thu Feb 27, 2020 at 08:35:41AM +0100, Landry Breuil wrote:
> > > > > On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> > > > > > Hi All
> > > > > >
> > > > > > After truly hard work, I am excited to write this email. The
> > > > > > following diff contains the 3 ports which are all dependent on each
> > > > > > other.
> > > > > >
> > > > > > - devel/py-sip
> > > > > > - x11/py-qt5
> > > > > > - x11/qt5
> > > > > >
> > > > > > Almost all consumers fixed. For some you will find patches on
> > > > > > ports@, for some, private maintainer emails are out. math/octave is
> > > > > > not
> > > > > > happy now but maybe an update helps. I'll take care of this monster.
> > > > > >
> > > > > > What's the news?
> > > > > >
> > > > > > - Most of the work was done in qtbase.
> > > > > > - The qtbase port comes with vulkan and zstd support by default
> > > > > > enabled.
> > > > > > I hope this will works fine with !{amd64,i386} arches.
> > > > > > - s/c++11/c++17/
> > > > > > - Qtdoc (docs) is broken again, I think I am not that bad again and
> > > > > > can
> > > > > > be fixed later.
> > > > > > - Bump all shred lib and cleanup SHARED_LIBS.
> > > > > > - Many cleaning jobs in the Makefiles.
> > > > > > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look
> > > > > > at this.
> > > > >
> > > > > can you explain why ? the MODPY_COMMENT bit should achieve the same in
> > > > > the current port.. unless i missed something ?
> > > >
> > > > I can't, this works for me. I'm not the py-{port} expert maybe there is
> > > > a smarter/cleaner way to do it.
> > >
> > > tried simplifying things to avoid the PFRAG.python3 & PLIST churn, but
> > > the resulting pyqt5 seems broken:
> > >
> > > [18:33] c64:~/mystuff/x11/py-qt5/ $python3
> > > >>> from PyQt5 import QtCore
> > > Traceback (most recent call last):
> > > File "", line 1, in
> > > ModuleNotFoundError: No module named 'PyQt5.sip'
> >
> > Looks like the py-qt5 diff misses "--sip-module PyQt5.sip --no-tools" or
> > whatever
> > it was that's needed..
>
> of course tried it but seems that option was only present in 5.11, ie now i
> get
>
> configure.py: error: no such option: --sip-module
>
Looking at
https://www.riverbankcomputing.com/static/Docs/PyQt5/installation.html
"It is not necessary to install the PyQt5.sip module before building
PyQt5 but it must be installed before PyQt5 can be used.
The module is built using setuptools and is available from the PyQt5-sip
project at PyPI. It uses setuptools as its build system and can be
installed by pip or you can also unpack the sdist and install it by
running its setup.py script."
So it looks like this should probably be a separate port ..
Re: UPDATE: Qt 5.13.2
On Sat, Feb 29, 2020 at 03:51:13PM +, Stuart Henderson wrote:
> On 2020/02/29 07:27, Landry Breuil wrote:
> > On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> > > On Thu Feb 27, 2020 at 08:35:41AM +0100, Landry Breuil wrote:
> > > > On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> > > > > Hi All
> > > > >
> > > > > After truly hard work, I am excited to write this email. The
> > > > > following diff contains the 3 ports which are all dependent on each
> > > > > other.
> > > > >
> > > > > - devel/py-sip
> > > > > - x11/py-qt5
> > > > > - x11/qt5
> > > > >
> > > > > Almost all consumers fixed. For some you will find patches on
> > > > > ports@, for some, private maintainer emails are out. math/octave is
> > > > > not
> > > > > happy now but maybe an update helps. I'll take care of this monster.
> > > > >
> > > > > What's the news?
> > > > >
> > > > > - Most of the work was done in qtbase.
> > > > > - The qtbase port comes with vulkan and zstd support by default
> > > > > enabled.
> > > > > I hope this will works fine with !{amd64,i386} arches.
> > > > > - s/c++11/c++17/
> > > > > - Qtdoc (docs) is broken again, I think I am not that bad again and
> > > > > can
> > > > > be fixed later.
> > > > > - Bump all shred lib and cleanup SHARED_LIBS.
> > > > > - Many cleaning jobs in the Makefiles.
> > > > > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at
> > > > > this.
> > > >
> > > > can you explain why ? the MODPY_COMMENT bit should achieve the same in
> > > > the current port.. unless i missed something ?
> > >
> > > I can't, this works for me. I'm not the py-{port} expert maybe there is
> > > a smarter/cleaner way to do it.
> >
> > tried simplifying things to avoid the PFRAG.python3 & PLIST churn, but
> > the resulting pyqt5 seems broken:
> >
> > [18:33] c64:~/mystuff/x11/py-qt5/ $python3
> > >>> from PyQt5 import QtCore
> > Traceback (most recent call last):
> > File "", line 1, in
> > ModuleNotFoundError: No module named 'PyQt5.sip'
>
> Looks like the py-qt5 diff misses "--sip-module PyQt5.sip --no-tools" or
> whatever
> it was that's needed..
of course tried it but seems that option was only present in 5.11, ie now i get
configure.py: error: no such option: --sip-module
Re: UPDATE: Qt 5.13.2
On 2020/02/29 07:27, Landry Breuil wrote:
> On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> > On Thu Feb 27, 2020 at 08:35:41AM +0100, Landry Breuil wrote:
> > > On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> > > > Hi All
> > > >
> > > > After truly hard work, I am excited to write this email. The
> > > > following diff contains the 3 ports which are all dependent on each
> > > > other.
> > > >
> > > > - devel/py-sip
> > > > - x11/py-qt5
> > > > - x11/qt5
> > > >
> > > > Almost all consumers fixed. For some you will find patches on
> > > > ports@, for some, private maintainer emails are out. math/octave is not
> > > > happy now but maybe an update helps. I'll take care of this monster.
> > > >
> > > > What's the news?
> > > >
> > > > - Most of the work was done in qtbase.
> > > > - The qtbase port comes with vulkan and zstd support by default enabled.
> > > > I hope this will works fine with !{amd64,i386} arches.
> > > > - s/c++11/c++17/
> > > > - Qtdoc (docs) is broken again, I think I am not that bad again and can
> > > > be fixed later.
> > > > - Bump all shred lib and cleanup SHARED_LIBS.
> > > > - Many cleaning jobs in the Makefiles.
> > > > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at
> > > > this.
> > >
> > > can you explain why ? the MODPY_COMMENT bit should achieve the same in
> > > the current port.. unless i missed something ?
> >
> > I can't, this works for me. I'm not the py-{port} expert maybe there is
> > a smarter/cleaner way to do it.
>
> tried simplifying things to avoid the PFRAG.python3 & PLIST churn, but
> the resulting pyqt5 seems broken:
>
> [18:33] c64:~/mystuff/x11/py-qt5/ $python3
> >>> from PyQt5 import QtCore
> Traceback (most recent call last):
> File "", line 1, in
> ModuleNotFoundError: No module named 'PyQt5.sip'
Looks like the py-qt5 diff misses "--sip-module PyQt5.sip --no-tools" or
whatever
it was that's needed..
Re: UPDATE: Qt 5.13.2
On Thu, Feb 27, 2020 at 08:48:55AM +0100, Rafael Sadowski wrote:
> On Thu Feb 27, 2020 at 08:35:41AM +0100, Landry Breuil wrote:
> > On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> > > Hi All
> > >
> > > After truly hard work, I am excited to write this email. The
> > > following diff contains the 3 ports which are all dependent on each
> > > other.
> > >
> > > - devel/py-sip
> > > - x11/py-qt5
> > > - x11/qt5
> > >
> > > Almost all consumers fixed. For some you will find patches on
> > > ports@, for some, private maintainer emails are out. math/octave is not
> > > happy now but maybe an update helps. I'll take care of this monster.
> > >
> > > What's the news?
> > >
> > > - Most of the work was done in qtbase.
> > > - The qtbase port comes with vulkan and zstd support by default enabled.
> > > I hope this will works fine with !{amd64,i386} arches.
> > > - s/c++11/c++17/
> > > - Qtdoc (docs) is broken again, I think I am not that bad again and can
> > > be fixed later.
> > > - Bump all shred lib and cleanup SHARED_LIBS.
> > > - Many cleaning jobs in the Makefiles.
> > > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at this.
> >
> > can you explain why ? the MODPY_COMMENT bit should achieve the same in
> > the current port.. unless i missed something ?
>
> I can't, this works for me. I'm not the py-{port} expert maybe there is
> a smarter/cleaner way to do it.
tried simplifying things to avoid the PFRAG.python3 & PLIST churn, but
the resulting pyqt5 seems broken:
[18:33] c64:~/mystuff/x11/py-qt5/ $python3
>>> from PyQt5 import QtCore
Traceback (most recent call last):
File "", line 1, in
ModuleNotFoundError: No module named 'PyQt5.sip'
i of course have py3-sip 4.19.19 installed. Will try to rebuild all the
py-* bits from scratch but that's definitely weird.
this failure leads to py-qscintilla failing:
[07:25] c64:/usr/ports/editors/py-qscintilla/ $env FLAVOR=python3 make
===> Configuring for QScintilla-2.11.4
Error: Unable to import PyQt5.QtCore. Make sure PyQt5 is installed.
Here's the pyqt diff i have, with only PLIST regenerated.
other than that, all the qt5 bits i built were fine (on amd64).
? pkg/PFRAG.python3
Index: Makefile
===
RCS file: /cvs/ports/x11/py-qt5/Makefile,v
retrieving revision 1.21
diff -u -r1.21 Makefile
--- Makefile19 Feb 2020 15:27:46 - 1.21
+++ Makefile29 Feb 2020 06:21:24 -
@@ -2,8 +2,8 @@
COMMENT= python Qt5 bindings
-V= 5.9.2
-DISTNAME= PyQt5_gpl-${V}
+V= 5.13.2
+DISTNAME= PyQt5-${V}
PKGNAME= py-qt5-${V}
CATEGORIES=x11 devel
@@ -14,31 +14,34 @@
# GPLv3 with exceptions for use with certain named open-source licenses
PERMIT_PACKAGE=Yes
-WANTLIB = ${COMPILER_LIBCXX} ${MODPY_WANTLIB}
-WANTLIB += dbus-1 glib-2.0 intl m pthread pulse pulse-mainloop-glib
-WANTLIB += GL Qt5Bluetooth Qt5Core Qt5DBus Qt5Designer
-WANTLIB += Qt5Gui Qt5Help Qt5Multimedia Qt5MultimediaWidgets Qt5Network
-WANTLIB += Qt5OpenGL Qt5Positioning Qt5PrintSupport Qt5Qml Qt5Quick
-WANTLIB += Qt5QuickWidgets Qt5Sensors Qt5SerialPort Qt5Sql Qt5Svg
-WANTLIB += Qt5Test Qt5WebChannel Qt5WebKit Qt5WebKitWidgets Qt5WebSockets
-WANTLIB += Qt5Widgets Qt5X11Extras Qt5Xml Qt5XmlPatterns
-WANTLIB += Qt5Location Qt5Nfc
+WANTLIB += ${COMPILER_LIBCXX} GL Qt5Bluetooth Qt5Core Qt5DBus
+WANTLIB += Qt5Designer Qt5Gui Qt5Help Qt5Location Qt5Multimedia
+WANTLIB += Qt5MultimediaWidgets Qt5Network Qt5NetworkAuth Qt5Nfc
+WANTLIB += Qt5OpenGL Qt5Positioning Qt5PositioningQuick Qt5PrintSupport
+WANTLIB += Qt5Qml Qt5Quick Qt5QuickWidgets Qt5RemoteObjects Qt5Sensors
+WANTLIB += Qt5SerialPort Qt5Sql Qt5Svg Qt5Test Qt5WebChannel Qt5WebKit
+WANTLIB += Qt5WebKitWidgets Qt5WebSockets Qt5Widgets Qt5X11Extras
+WANTLIB += Qt5Xml Qt5XmlPatterns dbus-1 glib-2.0 intl m pulse
+WANTLIB += pulse-mainloop-glib ${MODPY_WANTLIB}
FLAVORS = python3
FLAVOR ?=
BUILD_DEPENDS= ${RUN_DEPENDS}
-RUN_DEPENDS= devel/py-sip${MODPY_FLAVOR}>=4.19.1v0 \
+RUN_DEPENDS= devel/py-sip${MODPY_FLAVOR}>=4.19.19v0 \
x11/dbus-python${MODPY_FLAVOR}
+
LIB_DEPENDS= ${MODPY_LIB_DEPENDS} \
x11/qt5/qtconnectivity \
x11/qt5/qtdeclarative \
x11/qt5/qtlocation \
+ x11/qt5/qtnetworkauth \
x11/qt5/qtmultimedia \
x11/qt5/qtserialport \
x11/qt5/qtsensors \
x11/qt5/qtsvg \
x11/qt5/qttools \
+ x11/qt5/qtremoteobjects \
x11/qt5/qtwebchannel \
x11/qt5/qtwebkit \
x11/qt5/qtwebsockets \
@@ -47,11 +50,19 @@
NO_TEST= Yes
-MASTER_SITES= ${MASTER_SITE_SOURCEFORGE:=pyqt/}
+MASTER_SITES= https://www.riverbankcomputing.com/static/Downloads/PyQt5/${V}/
MODULES= lang/python \
x11/qt5
+.if !${FLAVOR:Mpython3}
Re: UPDATE: Qt 5.13.2
On Thu Feb 27, 2020 at 08:35:41AM +0100, Landry Breuil wrote:
> On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> > Hi All
> >
> > After truly hard work, I am excited to write this email. The
> > following diff contains the 3 ports which are all dependent on each
> > other.
> >
> > - devel/py-sip
> > - x11/py-qt5
> > - x11/qt5
> >
> > Almost all consumers fixed. For some you will find patches on
> > ports@, for some, private maintainer emails are out. math/octave is not
> > happy now but maybe an update helps. I'll take care of this monster.
> >
> > What's the news?
> >
> > - Most of the work was done in qtbase.
> > - The qtbase port comes with vulkan and zstd support by default enabled.
> > I hope this will works fine with !{amd64,i386} arches.
> > - s/c++11/c++17/
> > - Qtdoc (docs) is broken again, I think I am not that bad again and can
> > be fixed later.
> > - Bump all shred lib and cleanup SHARED_LIBS.
> > - Many cleaning jobs in the Makefiles.
> > - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at this.
>
> can you explain why ? the MODPY_COMMENT bit should achieve the same in
> the current port.. unless i missed something ?
I can't, this works for me. I'm not the py-{port} expert maybe there is
a smarter/cleaner way to do it.
>
> > - patch-qmake_generators_unix_unixmake_cpp
> > -- That was the biggest problem, at the p2k19 I decided to solve by:
> > "Transform /usr/ports/pobj/xxx/lib/libQt5Core.so into
> > -L/usr/ports/pobj/xxx/build-amd64/lib -lQt5Core" ... works!
>
> i suppose that lead to issues when linking qmake projects ? seems to be
> fine fixed this way, but im no linker expert.
It shouldn't be. This patch is for qmake code or respectively qmake
itself.
>
> > Looking forward for feedback.
>
> will build all this and my qt5 apps (ie mostly qgis & keepassxc) and
> dogfood all this on amd64, thanks for the hard work :)
keepassxc works fine here fine for over weeks.
Re: UPDATE: Qt 5.13.2
On Thu, Feb 27, 2020 at 07:59:32AM +0100, Rafael Sadowski wrote:
> Hi All
>
> After truly hard work, I am excited to write this email. The
> following diff contains the 3 ports which are all dependent on each
> other.
>
> - devel/py-sip
> - x11/py-qt5
> - x11/qt5
>
> Almost all consumers fixed. For some you will find patches on
> ports@, for some, private maintainer emails are out. math/octave is not
> happy now but maybe an update helps. I'll take care of this monster.
>
> What's the news?
>
> - Most of the work was done in qtbase.
> - The qtbase port comes with vulkan and zstd support by default enabled.
> I hope this will works fine with !{amd64,i386} arches.
> - s/c++11/c++17/
> - Qtdoc (docs) is broken again, I think I am not that bad again and can
> be fixed later.
> - Bump all shred lib and cleanup SHARED_LIBS.
> - Many cleaning jobs in the Makefiles.
> - New py-qt5/pkg/PFRAG.python3. Could the py- wizards take a look at this.
can you explain why ? the MODPY_COMMENT bit should achieve the same in
the current port.. unless i missed something ?
> - patch-qmake_generators_unix_unixmake_cpp
> -- That was the biggest problem, at the p2k19 I decided to solve by:
> "Transform /usr/ports/pobj/xxx/lib/libQt5Core.so into
> -L/usr/ports/pobj/xxx/build-amd64/lib -lQt5Core" ... works!
i suppose that lead to issues when linking qmake projects ? seems to be
fine fixed this way, but im no linker expert.
> Looking forward for feedback.
will build all this and my qt5 apps (ie mostly qgis & keepassxc) and
dogfood all this on amd64, thanks for the hard work :)
Landry
