[QGIS-Developer] Running unit tests without "make install"

[Thomas Larsen Wessel]

And a related question: I see that some of the tests create/show some
GUI-elements, even if just for fractions of a second. I have been both
typing and browsing while the tests have been running. Is that safe, or am
I potentially causing tests to fail?

On Tue, Jan 16, 2024 at 12:14 AM Thomas Larsen Wessel <mrvelle at gmail.com>
wrote:

> I mistakenly forgot to post the corresponding log (LastTest.log) before,
> but here it is:
> https://gist.github.com/velle/023e0cc6c3b5f134cdab3219cb9a8863
>
> Its running the test with the verbose flag now, but that will take an hour
> or so. And as far as I can tell it does not contain more information than
> the above, though it may be formatted in a more helpful way.
>
> On Mon, Jan 15, 2024 at 10:31 PM Nyall Dawson <nyall.dawson at gmail.com>
> wrote:
>
>>
>>
>> On Tue, 16 Jan 2024, 7:28 am Thomas Larsen Wessel, <mrvelle at gmail.com>
>> wrote:
>>
>>> Thanks for having a look at it :) I ran a complete test and saved the
>>> test output to this gist. Is that what you were looking for? Please tell me
>>> what other info you need.
>>>
>>> https://gist.github.com/velle/b195bee9a5d63e95ea7aa600b7833be1
>>>
>>
>> Try with "ctest -v" to get the full log of the tests
>>
>> Nyall
>>
>>
>>
>>>
>>> On Mon, Jan 15, 2024 at 7:54 AM Nyall Dawson <nyall.dawson at gmail.com>
>>> wrote:
>>>
>>>>
>>>>
>>>> On Mon, 15 Jan 2024, 3:42 pm Thomas Larsen Wessel, <mrvelle at gmail.com>
>>>> wrote:
>>>>
>>>>> When you guys run these tests locally, do they all pass?
>>>>>
>>>>> After some fiddling I have removed some errors, caused by incorrect
>>>>> build setup. But I still have  19 failing tests.
>>>>>
>>>>> 2 - checkGitStatus (Failed)
>>>>> 22 - ProcessingGrass7AlgorithmsRasterTestPt2 (Failed)
>>>>> 24 - ProcessingOtbAlgorithmsTest (Failed)
>>>>> 46 - test_core_compositionconverter (Failed)
>>>>> 108 - test_core_layoutpicture (Failed)
>>>>> 177 - test_core_settingsentry (Failed)
>>>>> 280 - test_gui_filedownloader (Subprocess aborted)
>>>>> 288 - test_gui_featurelistcombobox (Subprocess aborted)
>>>>> 298 - test_gui_queryresultwidget (Subprocess aborted)
>>>>> 302 - test_3d_3drendering (Failed)
>>>>> 309 - test_3d_mesh3drendering (Failed)
>>>>> 332 - test_provider_wcsprovider (Failed)
>>>>> 413 - PyQgsAnnotation (Failed)
>>>>> 489 - PyQgsExternalStorageWebDav (Failed)
>>>>> 490 - PyQgsExternalStorageAwsS3 (Failed)
>>>>> 600 - PyQgsMapCanvas (Failed)
>>>>> 601 - PyQgsMapCanvasAnnotationItem (Failed)
>>>>> 672 - PyQgsProcessExecutablePt1 (Failed)
>>>>> 673 - PyQgsProcessExecutablePt2 (Failed)
>>>>>
>>>>> I think I understand checkGitStatus and can safely ignore that one.
>>>>> But what about the rest? Is it "normal" that so many tests fail when
>>>>> running tests locally?
>>>>>
>>>>
>>>> That's more than I'd expect to fail on 22.04 -- at least, the main qgis
>>>> ci workflow uses 22.04 and most of these are passing !
>>>>
>>>> (See https://github.com/qgis/QGIS/blob/master/.ci/test_flaky.txt and
>>>> https://github.com/qgis/QGIS/blob/master/.ci/test_blocklist_qt5.txt
>>>> for lists of the tests skipped on the GitHub workflows)
>>>>
>>>>
>>>>> Would any of these tests give a different result if I first ran "make
>>>>> install"?
>>>>>
>>>>
>>>> Unlikely.
>>>>
>>>> I'd be interested to see a full log of the failures if you want to
>>>> upload it to a gist/Pastebin/ somewhere.
>>>>
>>>> Nyall
>>>>
>>>>
>>>>> Im running Ubuntu 22.04, and installed all dependencies via APT as
>>>>> advised in the qgis build docs. I built QGIS release-3_34 from source and
>>>>> am now trying to test that build, by running "ctest".
>>>>>
>>>>>
>>>>> On Mon, Jan 1, 2024 at 12:34 AM Thomas Larsen Wessel <
>>>>> mrvelle at gmail.com> wrote:
>>>>>
>>>>>> Thanks both of you :) And happy new year.
>>>>>>
>>>>>> On Sun, Dec 31, 2023 at 11:38 PM Nyall Dawson <nyall.dawson at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Mon, 1 Jan 2024, 6:36 am Thomas Larsen Wessel via QGIS-Developer,
>>>>>>> <qgis-developer at lists.osgeo.org> wrote:
>>>>>>>
>>>>>>>> The docs (
>>>>>>>> https://docs.qgis.org/3.28/en/docs/developers_guide/unittesting.html#run-your-tests)
>>>>>>>> specify: *make && make install && make test*.
>>>>>>>>
>>>>>>>> I would rather not run make install, since I have another instance
>>>>>>>> of QGIS installed via APT, and I worry running make install will somehow
>>>>>>>> make a mess and leave me without a working instance.
>>>>>>>>
>>>>>>>> But is it really necessary to run make install before make run? In
>>>>>>>> fact it seems more natural to me to test before installing!
>>>>>>>>
>>>>>>>
>>>>>>> Try running "ctest" from your build directory. There's no need to
>>>>>>> install if doing it this way. (It's how I do all of my development)
>>>>>>>
>>>>>>> Nyall
>>>>>>>
>>>>>>>>
>>>>>>>> Sincerely
>>>>>>>> _______________________________________________
>>>>>>>> QGIS-Developer mailing list
>>>>>>>> QGIS-Developer at lists.osgeo.org
>>>>>>>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>>>>>>>> Unsubscribe:
>>>>>>>> https://lists.osgeo.org/mailman/listinfo/qgis-developer
>>>>>>>>
>>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20240116/4f35e68f/attachment.htm>