[QGIS-Developer] Is there a recommended GDAL version for QGIS 3.34?

[Matthias Kuhn]

Hi Thomas,

I would recommend opening a pull request (flag it as draft if not yet sure
if it's ready for review) and see what feedback you get from CI.

You can also run a limited number of tests locally for a shorter feedback
loop during development / feedback for test results. Your time is probably
better spent developing than on getting a fully functional local test
environment, just make sure you don't actually commit any changed gpkg into
the pull request.

Cheers
Matthias

On Sat, Sep 7, 2024 at 10:36 AM Thomas Larsen Wessel via QGIS-Developer <
qgis-developer at lists.osgeo.org> wrote:

> Could someone please tell me if Im correct in thinking that the unittest
> on GitHub, CI for Qt5, will also fail because GDAL v. 3.4.1 updates
> timestamps of gpkg files, even when just reading them. And if it does not
> fail "checkGitStatus", then why not? What is different between my Ubuntu
> 22.04 and the Ubuntu 22.04 that runs tests on GitHub?
>
> I have more tests failing locally than just gitCheckStatus, but I chose
> this one as the first one to deal with. And I thought compiling against a
> more recent version of GDAL was the solution.
>
> If I've gone off track, or completely misunderstood something, then please
> give me a hint to get back on track :)
>
>
> On Tue, Sep 3, 2024 at 9:13 PM Thomas Larsen Wessel <mrvelle at gmail.com>
> wrote:
>
>> The instructions on the repo tell me, in rather strict wording, to run
>> the unit tests locally before creating a PR.
>>
>> The test checkGitStatus keeps failing locally. It does so because GDAL
>> updates the timestamp of gpkg files, even when only reading them. The
>> Internet told me that I needed to build against GDAL v. 3.4.2 or greater,
>> to not have these timestamps updated. So thats what I have spent a few
>> evenings trying to do. There are other tests that fail too, but this is the
>> one I chose to look at first.
>>
>> So it's quite surprising that you tell me CI for Qt5 uses GDAL 3.4.1.
>> Wouldn't that cause the automatic online test on github to fail? Or have I
>> misunderstood something? :)
>>
>> Sincerely, Thomas
>>
>>
>>
>>
>> On Tue, Sep 3, 2024 at 8:32 PM Even Rouault via QGIS-Developer <
>> qgis-developer at lists.osgeo.org> wrote:
>>
>>> For operational uses, I would say the latter is the better. For
>>> regression test purposes, some tests might perhaps need tweaking/skipping
>>> depending on the GDAL version. CI uses GDAL 3.4.1 for QT5 testing and 3.7.3
>>> for QT6 testing.
>>>
>>> > Background/motivation for asking: I have tried compiling different
>>> versions of QGIS against different versions of GDAL, but some unit tests
>>> keep causing segmentation faults. Could a segmentation fault be caused by a
>>> mismatch of version of a dependent library like GDAL?
>>>
>>> Nobody can guess that. You need to provide much more details: which
>>> test(s), stack trace when it crashes, etc.
>>>
>>> -- http://www.spatialys.com
>>> My software is free, but my time generally not.
>>>
>>> _______________________________________________
>>> 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
>>>
>> _______________________________________________
> 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/20240907/62d1f8f9/attachment.htm>