| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
Saves some typing and makes COIN integration easier.
Change-Id: Ic7b302157dd5b53388ccc7b89f70db5635bb3ba8
Reviewed-by: Christian Tismer <tismer@stackless.com>
|
|
|
|
|
|
|
|
|
| |
There is a small error that reports the wrong total number of changes.
Furthermore, the dependency of the six modure could be removed by a single line.
Change-Id: I510eab49ae3b9c5c36a7ae31a982334fda18567d
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
The testrunner.py script was generalisation to allow to select older build
versions. (This feature is not ready and in most cases it makes only sense
when comparing python 2 and 3.)
I had a wrong argument in “getcwd”. Fixed.
Change-Id: Ic948a5a6649535bd6ad3d23bbc846f03bde82de0
Reviewed-by: Simo Fält <simo.falt@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Testrunner did only run the pyside2 tests.
This new version also allows to specify the projects “shiboken2” and “pyside2-tools”.
Now multiple tests can be specified.
Interesting:
All shiboken2 tests succeed, all pyside2-tools tests fail.
Update: the pyside-tools tests work too, but the setup does not work
automatically, yet. A “make” must be run in the pyside2-tools directory,
and “pyside2-rcc -o example_rc.py example.qrc” must be run in order to
generate a resource file.
The tests are run according to a list. The default is currently
“—projects=shiboken2 pyside2”.
Change-Id: I69f08402de0956098aaca5553f98ce50cbf6eae5
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
|
|
|
|
|
|
|
|
| |
Python 2.X returns 'linux2' as platform for compatibility reasons regardless
of the kernel version whereas Python 3.3 onwards returns 'linux'.
Fixes blacklisting to work with newer versions of Python.
Change-Id: I1d3fd72feaa42bd5c89fa10a9a73474a92a6ab53
Reviewed-by: Christian Tismer <tismer@stackless.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The script was mistakenly reporting
FAIL : QtCore::bug_PYSIDE()
for
FAIL : QtCore::bug_PYSIDE-164()
Fix by extending the regular expression.
Change-Id: Id6ca4c75489bd577e25723efd62840f714e67ae4
Reviewed-by: Christian Tismer <tismer@stackless.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The testrunner module allowed only the last build to be tested.
This patch allows to specify an arbitrary test index.
The default is -1, which gives the normal behavior.
The argument ‘—buildno’ specifies which build is to be selected.
Note that multiple builds with the same directory are not valid, because the
latest entry always overwrites the older ones. But for different configurations
this makes sense.
The default runtime was set to 600 s.
Change-Id: Ib5b33a09dc0af7e36230ca4eba3c21c8970d8212
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
|
|
|
|
|
|
| |
Testrunner has even more variable texts to recognize.
We change the regex slightly so that it always succeeds.
Change-Id: Iac156592aac48afb5aea522540ae63c92ca2572a
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|
|
|
|
|
| |
Change-Id: I049929e2c4e24554866000df5daa063071060884
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are a couple of effects when running testrunner.py in real life.
Fix: It reports success when there are no tests found.
Fix and extend the recognition of relative paths.
Add a "getcwd" sub-command to get external access to the build folder.
The make command uses no longer "-C" for easier environ access.
Change-Id: Iee2f81ed762b9773021055d26f9b144c31acbe35
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
PySide is currently in an unfinished state. Therefore, running all the
tests will never return a success.
To make this more useful, this script is controlled by a blacklist,
in order to keep record of the known-to-fail tests.
On success: zero error status
On Failure: nonzero (raises ValueError)
The test parser uses a blacklist file that is compatible to standard
Qt blacklist files.
macOS, Ubuntu and windows (thanks to fkleint) are supported.
The blacklist has a feature where new configurations are learned.
It first removes old versions of the same tests, before it inserts the new ones.
Blacklisting is now optional.
We support now py2, py3, qt5, qt5.6 etc.
Some consideration about collapsing the data are added.
The parser has now a sub-command "test" that is mandatory.
Support for moving the build_dir to another location without loosing the tests so far.
Implement BPASS.
Support CTEST_OUTPUT_ON_FAILURE
Change-Id: If66d3cf1733a3b0c8a47fe9b8c6aec8ea430c699
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
|