| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
| |
Some of the new benchmarks in qmlbench require qtlottie and qtsvg to
run. Add them to the module_set of the qmlbench builder.
Change-Id: I62fcde6ca244419e7772f591e77b6131d064f432
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
| |
Change-Id: I78e78300791c623b27b1db661750cb9c88f82f07
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
| |
Change e1f258e8a59ee3175aa4e164c8392931910cc468 removed the code
path for the non-macOS platforms.
Change-Id: I5130b63488bf4d5498b880fe8c9f5f8e5b07fe6e
Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>
|
| |
|
|
|
|
|
|
|
| |
Previously test assumed that the binary is wrapped in the .app bundle and tried to execute with that path, which caused the "No such file or
directory" error.
This patch wraps the binary in Contents/MacOS/ and executes it.
Change-Id: Icdb2a99a65d3efc2de21301b329607e485be113f
Reviewed-by: Shyamnath Premnadh <Shyamnath.Premnadh@qt.io>
|
| |
|
|
|
|
|
|
| |
- Updated functions to use pathlib.Path for path manipulations and
directory changes
Change-Id: I5962a4173f2da5d4acd22f62171d837b4dcef0d3
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
| |
|
|
|
|
|
|
|
|
| |
- test_deploy now copies the entire project directory, not just .py
files. For the sample deployment this worked still because it did
not have any additional files, but gave a warning since the project
file was not copied.
Change-Id: I6f7594c9dbae969d3a8500dafafe8fc4085434cb
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, the javadoc doclet used to generate the API signature and
description of the Java APIs was using a 3rd party project under a
separate repo, that's not ideal. Now, the doclet is moved under qtqa.
Prior to this, it was expected that the binary of that doclet is
provided to the script manually. Now, it would be compiled and used
directly internally by the script. For that reason, we no longer need
the --doclet argument.
The expected output now is a single file describing the API of a given
module instead of a directory with multiple files representing various
classes in the API.
Additionally, the doclet was using old JDK APIs that are deprecated. It
used to require JDK 11, with this change, newer JDKs can be used because
we're using newer APIs. For that reason we no longer need the --javadoc
argument.
Pick-to: 6.10 6.9 6.8
Change-Id: Ifc1568a7a716d3eb740dba1ea3b5a114f2201ebe
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
|
| |
|
|
|
| |
Change-Id: I7fb0ab87f493040384e24493e482c757f43dcf2d
Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>
|
| |
|
|
|
|
|
|
|
| |
Treat the new header as if it were part of the copyright header.
Pick-to: 6.10
Task-number: QTBUG-135177
Change-Id: I6a21d0a966d05308a5d35641561c8c9287d8c03f
Reviewed-by: Cristian Maureira-Fredes <cristian.maureira-fredes@qt.io>
|
| |
|
|
|
| |
Change-Id: I7fefd981b5e0893dcb7a97ec322fa6005f834480
Reviewed-by: Samuli Piippo <samuli.piippo@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When no PySide was present, the too would error out:
testwheel.py", line 520, in <module>
VERSION[0] = path_version
TypeError: 'tuple' object does not support item assignment
This apparently resulted from some old code trying to determine
the version from the path.
Split out the code trying to find the root into a helper
function and reorder the code for clarity. Print out errors
to stderr.
Change-Id: I3f4e878cf3e3d09466180c65e92da98946ee6826
Reviewed-by: Shyamnath Premnadh <Shyamnath.Premnadh@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Pass the jenkins link where the wheels are stored using the
`--wheels-link` or the `-w`option and the script will automatically
download the wheels concurrently and install them in the virtual
environment.
- A number of 4 retries are done in case of a failure to download a
wheel. This could be due to a network issue, VPN disconnection, etc.
- Installing the wheels is done with the pip option of `--no-deps`
because of the random order in which the wheels are returned and
this could lead to a failure.
- A new requirements_testwheel.txt file is added.
- Add a check for virtual environment.
Change-Id: I38a098a21e93ffa8043d146cae19579851660987
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
| |
|
|
|
|
| |
Task-number: QTQAINFRA-6902
Change-Id: Ic97da89d2695d279d1b24918d9dcf07e84c208c4
Reviewed-by: Ville-Pekka Karhu <ville-pekka.karhu@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
| |
This sub-package was moved to the parent to allow using some internal
APIs from the parent package.
See 523cf066285a648b10b277bb1b3b2c7b9d66d464.
Pick-to: 6.9 6.8
Change-Id: I9328ffea212df207e4aaa381cc5f51aa434d08c0
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
|
| |
|
|
|
|
|
|
| |
Add bumping version from android examples in qtdeclarative
Change-Id: Icf2b269fc645b72d3e229a146f38c363b7d4f2da
Reviewed-by: Olli Vuolteenaho <olli.vuolteenaho@qt.io>
Reviewed-by: Akseli Salovaara <akseli.salovaara@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a branching event occurs, this script now also removes Code-Review
+2 votes from changes on the dev branch which do not target the new
latest stable branch. This is to force a re-evaluation of the change
to determine if the change should be pushed through to the new FF stable
branch or if the change can simply merge normally to dev for the next
stable release.
This change also fixes a minor, but breaking bug in the original script
where the regex used to parse branches was incorrectly used.
Change-Id: Idadec12ca8d958f7d86e83bfeaaa123d121a7bab
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
| |
- Updated version check to 6.6.0, for PySide6 Addon wheels.
Change-Id: Idc3e886c7161707bef38e9b116e33b1705a86222
Reviewed-by: Shyamnath Premnadh <Shyamnath.Premnadh@qt.io>
|
| |
|
|
|
|
|
| |
qtqa will follow normal qt branching scheme
Change-Id: Iefeaacdee07716e94d005786249c07195f8b654c
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
The __all__ variable has been made lazy to compute it on-demand.
When using the variable normally like module.__all__ everything
is fine.
Direct access to a modules __dict__ circumvents every enhancement
made through __getattr__, because __getitem__ is directly used.
Change-Id: I3b930685098391ccff569003bd2b585eda8f5a26
Pick-to: master
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The script uses javadoc to generate java files that contain only
the API description, and its documentation. The script looks for
various Java sources under the qt5 super repo path and generates API
description for specific package names provided here, these pre-defined
Qt for Android package names. This process is done for the previous Qt
release and it's committed as a base to-be-ignored commit. The same is
done for Qt release being reviewed and a commit is created on top.
The script can push changes to Gerrit, and automatically abandon the
base commits. Also, it can update existing API review patches, given
that the local git branch that was used to push before still exists with
valid commits.
Optionally a Jira task number can be provided.
Fixes: QTBUG-126352
Change-Id: Ib5cc7c560e320dbda9e9df5104c06161dccd7561
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
|
| |
|
|
|
|
| |
Pick-to: master
Change-Id: I22d946d24c5edfbd751f54084f1fce67368ede6c
Reviewed-by: Shyamnath Premnadh <Shyamnath.Premnadh@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Add graphicsApi tag with value parsed out from qmlbench result json
- Push data to new bucket. Different tag values will be interpreted
as a new series and therefore those will always be writing these
new data points. The only way to update that tag value is to write
the new series and then delete the old.
Before taking this change in use, copy earlier bucket content
and while copying add new tag with default value to all datapoints.
Change-Id: I3a69c081b9cfd587cbd0eb4af6b234a92269aa3e
Reviewed-by: Audun Sutterud <audun.sutterud@qt.io>
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
| |
Add version bump for qtbase/tests/auto/tools/rcc/data/legal/rcc_legal.cpp
Change-Id: I71d2a27a1e4d8d2e93a0d13fd31a661d7145dcf6
Reviewed-by: Akseli Salovaara <akseli.salovaara@qt.io>
|
| |
|
|
|
|
|
|
|
| |
Some supported arguments were omitted.
Move usage() function after copyright header, now that the latter is
lightweight.
Change-Id: I94634ce39c114cdb0af0bd750ce8a3ec33b4c0f5
Reviewed-by: Lucie Gerard <lucie.gerard@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
| |
It has been decided that bots should have their own repositories. See
QTQAINFRA-6429 for information about this. The runner sources have been
copied to qtbots/qtcorebenchmarkrunner. This commit removes the original
source files.
Task-number: QTQAINFRA-6470
Change-Id: Ib779b6ee23b828c4e6a9d1c84908b6722e4e9299
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Because QT_LICENSE_ID starts with QT, like many defines (values within
a parameter, rather than the name of a parameter) it wasn't recognized
as the name of a CMake function parameter. Complicate the logic of
__is_param_name() a little bit more to cope with that. (It otherwise
caused the parameter to be included in the value of a preceding
PRIVATE_HEADER_FILTERS, for which it was an unsuitable value.)
Change-Id: I1788d8f80796b77018bb043b99b7f38e07a87a72
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
|
|
|
|
|
|
|
| |
This functionality has been implemented as a plugin
for Cherry-Pick Bot for some time now, living at
qtbots/qtcherrypickbot::plugin_bots/jira_closer
Change-Id: Ib63789c5d576bdea4f425611051aa03134142bd7
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
| |
this bot has moved to its own repo at qtbots/qtcherrypickbot
Task-number: QTQAINFRA-6258
Change-Id: I2cd2e4e8b1db3ba61a23c2c640ba7ed4bb0cb5af
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the event that a relation chain is integrated together, but one
change in the chain is not staged with the rest of the chain,
it is possible that any changes higher in the chain than the unmerged
one will skip more than just the missing change to find a parent.
This occurs under heavy load when cherry-picks for changes lower on the
relation chain on the target branch do not yet exist when walking
the relation chain for higher changes.
This fix introduces a check for each potential parent to see if it is
merged on the source branch and has a pick expected to the target
branch. If so, a 1 minute listener will be set up to pick the change
when the parent pick is also created. If the listener expires, the
parent will be re-checked for safety, and then skipped if it still does
not exist. This process repeats until a parent is found or the chain
is exhausted, at which time the change will be picked to the target
branch HEAD as a last resort. A message is also posted to the
created cherry-pick with references to all skipped parents.
Fixes: QTQAINFRA-6455
Change-Id: I454ae99ac46c6f5f0bb9dde6651361cd30e9e993
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch introduces a new footer, "Reopens", which reopens a JIRA
issue upon merge of the change. Reopens and Fixes targeting the same
jira issue should not be specified in the same commit. However, if This
occurs, the bot will prioritize the Fixes footer as a closing action.
In order to allow the use of Fixes: to close an issue after it has been
reopened by the bot or manually, the bot will now check issue comments
to determine if the issue has been previously closed by the currently
merging change ID. If so, the bot will ignore the close action.
In practice, this means that cherry-picks which include a Fixes footer
will not re-close an issue, but a new change with a fresh Change ID
will close the issue with the Fixes footer as expected.
The bot will also now automatically tag relevant jira issues
with the commit of a revert, if the commit message contains
the standard revert message "This reverts commit ([0-9a-f]{40})".
When a revert is detected, the bot will also tag any JIRA issues
tagged in the original reverted commit message footers.
Fixes: QTQAINFRA-6248
Change-Id: I92def4927516fbab5e520cd23d34032b3dc4ade7
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Quite a few cases are not expected to fail. These are now handled
in a minimal way to at least avoid a crash.
Further, the jira plugin will now intentionally throw an error
on startup if JIRA is not reachable, as this is critical to general
functionality.
Fixes: QTQAINFRA-6313
Change-Id: I51005b552ea07fd4a3718326e5da9277baf709a3
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
API specific: If an issue is queried and either does not have
a Fixes: footer, or querying that issue results in no fix versions,
the bot will now attempt to find the fix version from a a default
QTBUG query. This only applies if the change is in the qt/ namespace.
This does not functionally alter which fix version should be applied
to the change, but allows for guessing fix versions via the API for
changes which do not specify a relationship to a Jira issue.
Fixes: QTQAINFRA-6312
Change-Id: I6237ca4fa79416afad515f39d8bb104290a6391f
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch changes the behavior of the cherry-pick bot to skip branches
that are already closed for new changes, instead of halting upon
encountering such. This behavoir change comes as a result of the move
to the waterfall model, where the pick bot is expected to pick targets
in a sequential manner rather than simultaneously. This works around
the possible scenario where a change on dev is merged with valid pick
targets, but during the waterfall process, one of those targets is
closed. Now, the bot will still post a warning message about the closed
branch, but then continue to pick to the next target regardless.
Fixes: QTQAINFRA-6287
Fixes: QTQAINFRA-6288
Change-Id: I91d9e066b4bcba1763049272fdf26c3b267d2688
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
| |
When a cross-repo cherry pick cannot be completed, the URL
presented to the user is incorrect, and always references the original
project, not the target project.
Fixes: QTQAINFRA-6286
Change-Id: Ib29ffd7f9ac9988717e7ab2b105fa136d4a84b0e
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
| |
Bot now lives at qtbots/qtbinarysizebot
Task-number: QTQAINFRA-6459
Change-Id: Id8de1c0b2df82329255c95240fb6131045139fdd
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Bugs fixed:
- Startup
- Too fast COIN access
- Add retry for COIN access
- Incorrect urllib return code handling
- Incorrect InfluxDB pull query result
Change-Id: I1332983c4198c5ad12e6b0cc1b1ae14a5128494c
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
|
|
|
| |
Discussion of some headers, in the qttools review for 6.8, that
shouldn't have been public (but were) brought to light that syncqt
doesn't generate includes for headers in SOURCES accessed via a ../
path. So match that here.
Task-number: QTBUG-125859
Change-Id: Ie9381478010d62bc7748d02853c72f1485f8364c
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
|
|
|
|
| |
Task-number: QTBUG-122999
Change-Id: I84e75458582b2c42351f3048c08119140b32acf1
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
| |
The syntax --prefix=PATH is no longer valid. This should fix
QTQAINFRA-6353.
Change-Id: Ib42dbdb758a81405a58fcc177f562c9e09779f18
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
|
| |
This copes with qtgraphs' somewhat more complex CMake config.
Task-number: QTQAINFRA-6317
Change-Id: Iea99c8ed82e7bdd6afff6eadf636c216a18f96b0
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
|
| |
|
|
|
|
|
| |
Some of these errors can be serious and we should not attempt a recovery.
Change-Id: Ia23bbb019146c437e47ab0e5776a8cd9a31f59ee
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
| |
Pass the target explicitly to cmake. This fixes QTQAINFRA-6292.
Change-Id: Iba8cb01833886cd4e648bb45433c2fba1fb397f8
Reviewed-by: Dimitrios Apostolou <jimis@qt.io>
|
| |
|
|
|
|
|
|
| |
Fixes qtgraphs CMakeLists.txt scanning.
Fixes: QTQAINFRA-6317
Change-Id: I24fae6c70ecb214c02d87f290fc155bbb2950d9a
Reviewed-by: Mate Barany <mate.barany@qt.io>
|
| |
|
|
|
|
|
|
|
|
| |
Previously the last header from each CMakeLists.txt was missing a
trailing newline, causing it to be glued to the first header of the
next, when several were processed together.
Task-number: QTQAINFRA-6317
Change-Id: If967be85a8642e7fb38d38f8af1f5df5ddd3f299
Reviewed-by: Mate Barany <mate.barany@qt.io>
|
| |
|
|
|
|
|
|
| |
Recently reminded (by ./modunite) of this nicer pattern, I went
looking for more places to apply it.
Change-Id: Ic14399d8d7afc91d4e339f237f543da1b35f41d6
Reviewed-by: Mate Barany <mate.barany@qt.io>
|
| |
|
|
|
|
|
|
|
|
| |
- separate tools into ESSENTIAL_TOOLS and ADDONS_TOOLS
- ADDONS_TOOLS are tested only for AddOns wheel
Pick-to: master
Change-Id: I40fb978450af18c34c868adf3a4fa6b37ab18561
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Adrian Herrmann <adrian.herrmann@qt.io>
|
| |
|
|
|
|
| |
Pick-to: master
Change-Id: I72e7fa83a4b0810179ffff25969334366c63f091
Reviewed-by: Adrian Herrmann <adrian.herrmann@qt.io>
|
| |
|
|
|
|
|
| |
Pick-to: master
Change-Id: Iaee4f37ebf982a5d7648690979ddd14300f70386
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Adrian Herrmann <adrian.herrmann@qt.io>
|
| |
|
|
|
|
|
|
| |
Explicit termination is necessary after the cancellation of an
asyncio.subprocess.Process.wait().
Change-Id: I205352008f492cdcd5a741d3ae83570ff5cb7d0d
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|
| |
|
|
|
|
|
| |
Match-statements are error-prone. Use simple if-statements instead.
Change-Id: Ib287ade80187585142b2aa7ab39ea5c7bf3605dd
Reviewed-by: Daniel Smith <daniel.smith@qt.io>
|