The Python provisioning on macOS ARM hosts is done multiple times
across separate targets.
This patch moves the code into a common script 'macos/python-arm.sh'.
As a drive-by, we bump the version of all targets to 3.12.11, whch is
the same as the one used in the macOS 15 ARM host.
Pick-to: 6.10 6.8
Change-Id: I8c1f6f7d635fe47b9faa34dbe6988e89252dc2c9
Reviewed-by: Artem Dyomin <artem.dyomin@qt.io>
(cherry picked from commit f0b54a51f7)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
On macOS hosts we are using external mirrors for installing with pyenv.
We should instead pull the necessary files from our local mirror.
Pick-to: 6.10 6.8
Change-Id: I6ba67dfd7cd37809164c7e2c4bcf89d07d45b1b2
Reviewed-by: Artem Dyomin <artem.dyomin@qt.io>
(cherry picked from commit bb77943c0f)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Commit d6918d7ac5 broke init-repository if
you did not pass an --alternates option.
Pick-to: 6.10 6.8
Task-number: QTBUG-142712
Fixes: QTBUG-142744
Change-Id: Ia727b420a34a46e52f0f05e549aabe2c975c6d7e
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
(cherry picked from commit 6bd3b23b6c)
When using --alternates with a relative path, nested submodules
(e.g., qtdeclarative/tests/auto/qml/ecmascripttests/test262) failed
because the alternates path wasn't adjusted for the changed working
directory depth. The --alternates option worked correctly with absolute
paths.
Fix by converting the passed alternates path to an absolute path.
Pick-to: 6.10 6.8
Fixes: QTBUG-142712
Change-Id: Ief7a87768d9ff17bb2832ae363f573ecae58d5ee
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
(cherry picked from commit d6918d7ac5)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
installing libc++-20-dev and libgstreamer1.0-dev on the same system
causes problems:
* using the llvm apt repo, it will uninstall libgstreamer1.0-dev
* using ubuntu's repo, it will break gstreamer's pkg-config integration
reason is that libgstreamer1.0-dev depends on libunwind-dev, but llvm's
libunwind-20-dev does not provide the pkg-config files.
as temporary workaround we basically revert
e5b56d1d37.
Change-Id: Ib4626d17339256d56c4764e3167f4e9fe4b9c909
Reviewed-by: Tero Heikkinen <tero.heikkinen@qt.io>
Reviewed-by: Liang Qi <liang.qi@qt.io>
(cherry picked from commit e9a003f0c3)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
This helps with the test stability when code.qt.io is down or slow.
Pick-to: 6.10
Change-Id: I0bdaaa06772a94e73cb33d3f827e4e327123b60a
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
(cherry picked from commit 160e6135bb)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Renamed files since it leads to misunderstanding due to their index.
Our CI system sees only two first digits, so 100 is seen as 10
and executed with other 10s which is misleading.
Pick-to: 6.10 6.8
Change-Id: Iabb548c622d71df8a3593cc171b7e0c2e2ae0287
Reviewed-by: Artem Dyomin <artem.dyomin@qt.io>
(cherry picked from commit b276b549f7)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
Updating for raising the compiler requirements.
Also adds Ubuntu 22 to test with GCC 11
Since Qt Creator branches can use the same Qt branches
that define these platforms, keep one RHEL 8 job
for Qt Creator 18. Mark it in the features, so the
Qt Creator 19+ configuration can filter these out.
Pick-to: 6.10
Change-Id: Ib771f473b6ed3ce42c6191647da7d473b64c2316
Reviewed-by: Jukka Jokiniva <jukka.jokiniva@qt.io>
(cherry picked from commit dcd07cc2ab)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
- need to trigger an update for enabling changes required
for merging windows provisioning.
Change-Id: I8b5804c2bda2e4f4f4f204b0db95232583ac57e1
Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
The installer from Microsoft which is used for ARM machines has a
different default installation path compared to the installer from
Oracle that is used for x64 machines, but the provisioning script did
not take it into consideration.
As a result, the JAVA_HOME variable was pointing to a non-existing dir
on Windows ARM machines.
An attempt to adjust the installation path using TARGETDIR or INSTALLDIR
properties did not work, so this patch simply hardcodes the path that is
used by the current ARM installer.
Amends b3f5b4fd64 that added Java
provisioning to Windows ARM.
Task-number: QTBUG-138968
Pick-to: 6.10 6.8
Change-Id: I2abdc07195a9c633340c2d7b21deaa80762aee67
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
- Rename the sha256 variable to sha1, because that's the actual
algorithm that was used to calculate the checksum.
- Use the `version` variable when constructing `externalUrl`.
Amends 0ac4df3e52.
Change-Id: I79d38fdf5033dce7d0b04c2d8788767b2325895f
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
It is required for qtopenapi.
The patch adds the maven provisioning to only those Windows
configurations that are currently used in qtopenapi.
This commit amends 0ac4df3e52 which
provisioned the same files for Linux and macOS.
Task-number: QTBUG-138968
Change-Id: I8ec903959c58185a5c46bdb7c13f178bdc10a109
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
- Add Windows configurations that use MSVC
- Provision openapi generator on the relevant platforms
- golang is already provisioned on all these platforms
The patch does not enable mingw and llvm configs, because
we need to build zlib using mingw first.
It also does not enable x64_64->arm64 cross-compile configs,
because they have a dependency on the mingw config.
Task-number: QTBUG-138968
Change-Id: I4fe35d149a1f9bfa93c1a128e5910b33aad536d4
Reviewed-by: Tatiana Borisova <tatiana.borisova@qt.io>
Place the downloaded file into the new directory inside C:\Utils and
add it to the PATH, because that's what the qtopenapi build scripts
expect.
Task-number: QTBUG-138968
Change-Id: I113b0095074cc8f0dc0470daf54fa7f22ee22719
Reviewed-by: Toni Saario <toni.saario@qt.io>
That is required for qtopenapi builds that would be enabled in the
follow-up patch.
The zlib library is already built after commit
97e80c1538. That commit also sets the
ZLIB_PATH_{arch} environment variables. So, we only need to set
ZLIB_ROOT to point to the proper zlib installation.
This patch does not cover llvm and mingw configurations, because we
need to builds zlib using these compilers first.
Task-number: QTBUG-138968
Change-Id: I5d5ad754d13bef9fc6184b41823a42d98863991a
Reviewed-by: Toni Saario <toni.saario@qt.io>
libuuid is needed for it. RHEL 9.0 already has it in its package list,
but it was missing for the later RHEL 9 versions
Pick-to: 6.10
Change-Id: Ibd052438c534e5811d2f0b0ee848808227e795f9
Reviewed-by: Marcus Tillmanns <marcus.tillmanns@qt.io>
This currently flags 294 occurences in Qt Core as errors, practically
all of them related to the implemenation of QDataStream related
functionality.
I think this generally at the wrong level, certainly for Qt Core,
but also for user code: operator>> is (besides operator<<)
/the/ main reason to use the QDataStream class, so if we consider
this risky, QDataStream itself should not be used, i.e. it would
be better to flagging code that creates QDataStream objects not
every use of operator>> of such objects.
Change-Id: If758bde3a77f64cc8c06f8328459dfcd2d94dc8f
Reviewed-by: Matti Paaso <matti.paaso@qt.io>
To all linux, macOS and Windows platforms.
Task-number: QTQAINFRA-7554
Change-Id: I677440fb9faf3f3d36744b59f91f0036449b639b
Reviewed-by: Jukka Jokiniva <jukka.jokiniva@qt.io>
RHEL-9.6 will be run in parallel with RHEL-9.4 with insignificant tests
until RHEL-9.6 is confirmed stable, then it will replace RHEL-9.4.
[CI Platforms]
Task-number: QTQAINFRA-7528
Change-Id: I5fe6bfe6f57f1dfb7ec4aeb05560dd7f01fc7f2f
Reviewed-by: Tony Sarajärvi <tony.sarajarvi@qt.io>
Reviewed-by: Tero Heikkinen <tero.heikkinen@qt.io>
openSSL 3.2.0 or newer required for libcurl in openSUSE 16.0
This solution has possibility to leave default openSSL
version into use for others, and openSUSE 16.0 (or others)
can move to a newer version if needed.
There are currently two tests failing in this update,
but as openSUSE 16.0 is non-blocking in the CI, it
doesn't get blocked.
Tickets created to fix those:
- QTBUG-141722
- QTBUG-141723
Task-number: QTQAINFRA-7502
Task-number: QTQAINFRA-7455
Task-number: QTQAINFRA-7452
Change-Id: Ie9995a3afcdb09d091d75651528a94f71f5ece2a
Reviewed-by: Elias Toivola <elias.toivola@qt.io>
Reviewed-by: Simo Fält <simo.falt@qt.io>
Notes:
- GCC 15
- Python 3.13
- MimerSQL with openSSL 3
- CINetworkTest skipped until newer ICU 77 version available
- gstreamer-plugin-openh264 not available
Repository names have changed:
- Old: /etc/zypp/repos.d/repo-oss.repo
- New: /etc/zypp/repos.d/openSUSE:repo-oss.repo
Not installed by default, therefore no need to disable anymore:
- balooctl
- ntp
Task-number: QTQAINFRA-7455
Task-number: QTQAINFRA-7452
Change-Id: I2af7f828e07d31d3622cf46e819318c71d28783e
Reviewed-by: Simo Fält <simo.falt@qt.io>
Reviewed-by: Elias Toivola <elias.toivola@qt.io>
This is an unfortunate combination of clazy-function-args-by-ref and
clazy-function-args-by-value.
function-args-by-ref fixes are almost always safe to apply. They
actually may save an unbounded amount of cycles, and there are typically
only a few of them.
On the other hand, function-args-by-value is a mixed bag: There are a
lot of them and the expected gain is small: It's typically trading one
or two register for one register and a "hot" location on the stack.
Especially when passing down such objects to a through a sequence of
function calls this is not even necessarily a clear win on all
architecture.
In any case it should be considered that especially in the Qt context
changing signatures of public functions is not cheap, if possible
at all.
Change-Id: I281ea086fb15e0a045316591d9af620b5e730f4a
Reviewed-by: Matti Paaso <matti.paaso@qt.io>
Reviewed-by: Alex Blasche <alexander.blasche@qt.io>
The binaries will be stored to ci-files02 after successful integrations
and the testing/release process will use those binaries.
Currently we can disable tests, as there are no unit tests in the
plugin itself.
Task-number: QTQAINFRA-7390
Change-Id: I7777ee47eed845022d67447a9389ab6943d69bc9
Reviewed-by: Tero Heikkinen <tero.heikkinen@qt.io>
