Reasons:
* Port build shouldn't use cross-compilation mode
because it buids for the same architecture.
* RUSTFLAGS isn't passed to the build of all or some Rust ports
in the cross-compilation mode.
Essential changes:
* Remove the CARGO_BUILD_TARGET make variable.
* Remove CARGO_BUILD_TARGET and CARGO_TARGET_* from CARGO_ENV.
* Update many ports that used CARGO_BUILD_TARGET.
* Build paths now don't include the architecture triplet part.
Immadiate benefits:
* Unbreak build on architectures like i386 that sometimes require
special RUSTFLAGS that were not effective in the cross-compilation
mode.
Ports that really need cross-compilation for some reason should
enable it on case-by-case basis. Example: net-p2p/cncli
(Not sure why does net-p2p/cncli actually need it, but it only
builds with these options.)
PR: 280305
Approved by: rust@FreeBSD.org (maintainer's timeout; 98 days)
PyQt5:
- devel/py-qt5-pyqt: Update to 5.15.11
- graphics/py-qt5-3d: Add port
- net/py-qt5-networkauth: Update to 5.15.6
- www/py-qt5-webengine: Update to 5.15.7
- x11-toolkits/py-qt5-chart: Update to 5.15.7
- x11-toolkits/py-qt5-datavis3d: Add port
PyQt6:
- devel/py-qt6-pyqt: Update to 6.7.1
Build tools:
- devel/sip: Update to 6.8.6
- devel/py-qt5-sip: Update to 12.15.0
- devel/py-qt6-sip: Update to 13.8.0
- devel/py-qtbuilder: Update to 1.16.4
The build tools are now all licensed BSD2CLAUSE and have been converted
to PEP517.
make(1) has a built-in variable MACHINE_ARCH that defaults to the value
of sysctl hw.machine_arch. This is not correct when building packages
under qemu. In that case poudriere sets MACHINE_ARCH in its make.conf.
Since commit d7511b9d0074 make.conf is not picked up by make during
build so pass MACHINE_ARCH through the environment.
This affects building kmod ports under qemu.
This is a workaround for make not looking at UNAME_p when setting
MACHINE_ARCH. It does look at UNAME_m when setting MACHINE.
PR: 281674
Reported by: jrtc27
Exp-run by: antoine
Approved by: portmgr (antoine)
As a patch release, Qt 6.7.3 does not introduce new features but
contains more than 400 bug fixes, security updates, and other
improvements to the top of the Qt 6.7.2 release. See more information
about the most important changes and bug fixes from Qt 6.7.3 release
note.
PySide6: Update to 6.7.3
New port: devel/qt6-grpc
Qt GRPC is the client-side implementation that, together with the Qt
Protobuf module, allows the definition of messages and services in
.proto files, which then use the provided code generators to generate
client code that allows the access for fields and gRPC services in the
Qt framework. The code generated by Qt GRPC enables the client-side to
communicate with a gRPC server (regardless of whether the server uses
Qt) by sending calls or streaming messages.
Announcement: https://www.qt.io/blog/qt-6.7.3-released
Release note: https://code.qt.io/cgit/qt/qtreleasenotes.git/about/qt/6.7.3/release-note.md
Mk/Uses/gstreamer.mk:
- Sort, fix whitespace issues and typos
- Promote several common variables to improve maintainer QOL
* There's now no need to set manually set PORTVERSION nor
SOVERSION in several Makefiles
* This will help to prevent not updating some components
* A note has been added, however, to remind maintainers
to remake distinfo for certain ports
- Improve documentation
multimedia/gstreamer1-plugins-webrtc:
- New port split from multimedia/gstreamer1-plugins-bad
- multimedia/gstreamer1-plugins-rust has been updated
to reflect this change
net/gstreamer1-plugins-sctp:
- New port split from multimedia/gstreamer1-plugins-bad
x11-toolkits/gstreamer1-plugins-qt5:
- Moved here from graphics/gstreamer1-plugins-qt to
align with its new Qt6 counterpart
x11-toolkits/gstreamer1-plugins-qt6:
- New port: GStreamer Qt6 QML videosink plugin
*/*:
- Improve Makefile order and formatting
- Sorry for the repo churn, but not fixing this would just lead to more
poorly formatted gstreamer ports in the future via copypasta
https://gstreamer.freedesktop.org/releases/1.24/#1.24.8
PR: 278914
Reported by: vvd
When using USE_GITHUB=nodefault, we don't need to add GH to
MASTER_SITES, as the github urls that are needed will be added later.
Adding GH here means we'll add a generic GitHub URL to the fetch list,
in the form of github.com/<portname>/<portname>/.../<dist> which will
not work. When used with USE_GITLAB the framework will add this bogus
URL before the correct gitlab url. It does not prevent the distfile to
be fetched from gitlab, but it makes everyone try and fetch from a bogus
url before trying the correct url.
PR: 279816
When using USE_GITHUB=nodefault, we don't need to add GH to
MASTER_SITES, as the github urls that are needed will be added later.
Adding GH here means we'll add a generic GitHub URL to the fetch list,
in the form of github.com/<portname>/<portname>/.../<dist> which will
not work. When used with USE_GITLAB the framework will add this bogus
URL before the correct gitlab url. It does not prevent the distfile to
be fetched from gitlab, but it makes everyone try and fetch from a bogus
url before trying the correct url.
This reverts commit 0d1aba712a301ddb1e8b7ccbf3acc721dc3c90e8.
A recently reported bug 281360 might suggests that Samba 4.19 does not
work with Time Machine and might result in data loss.
Let's revert the default version bump for now until we have more time to
plan how we are going to deal with issue. A potential solution would be
to disable Time Machine support in 4.19 for now.
PR: 280769 281360
Reported by: theraven
Approved by: samba (0mp)
