[Qgis-user] bug in 3.10.2 (after upgrading from 3.10.1) with NULL being lost in PostGIS

Alessandro Pasotti apasotti at gmail.com
Tue Feb 4 12:35:59 PST 2020


Hi Jason,

the quickest way to test current master for your case is to unzip the
latest build artifact for windows 64bits (it's a zip inside a zip) on your
PC and launch the qgis.exe binary you find inside.
No installation is required and it's completely standalone, there is no
Python but you don't need it for your test case.

On every commit the artifact is automatically built:
https://github.com/qgis/QGIS/actions?query=workflow%3A%22Windows+cross+build+with+MXE%22

The patches are very recent and they will be backported to 3.10 but I'll do
that later this week.

Let me know how it goes.


On Tue, Feb 4, 2020 at 8:47 PM Jason Carlson <jason at starlandcounty.com>
wrote:

> Hi Alessandro,
>
> Thanks for your response.
>
> I've only used the standalone installers so it took me a bit to figure out
> how to start QGIS when using the OSGeo4W installer. My standalone was
> version 3.10.2-0 and in the OSGeo4W installer I see a 3.10.2-2 as well as
> the development version of 3.11.0-72.
>
> Wasn't sure which one you meant. If you meant the 3.11.0-72 then the issue
> is gone. If you had meant the 3.10.2-2 branch then it is still there but I
> discovered more stuff that may help but won't explain until I confirm the
> version I was to test against.
>
> Jason
>
> On Tue, Feb 4, 2020 at 9:05 AM Alessandro Pasotti <apasotti at gmail.com>
> wrote:
>
>>
>>
>> On Tue, Feb 4, 2020 at 5:02 PM Jason Carlson <jason at starlandcounty.com>
>> wrote:
>>
>>> Can anyone else confirm this before I submit it?
>>>
>>
>>
>> Can you please check latest master builds first? I've done a couple of
>> fixes in the past few days that might have an impact on this issue.
>>
>> If the issue is still there, please file a ticket and attach a small SQL
>> dump and a QGS project to reproduce the issue.
>>
>>
>>
>>
>>
>>> After I upgraded from 3.10.1 to 3.10.2 I noticed an issue that it no
>>> longer handles null values correctly when updating records in PostGIS.
>>>
>>> I rolled back to version 3.10.1 and the issue goes away.
>>>
>>> Here is more detail:
>>> I have some PostgreSQL tables that have fields that can be NULL or text
>>> but I have check constraint in place saying it has to be NULL or one of a
>>> list of specific strings.
>>>
>>> In 3.10.1 if I go to edit a record that has some fields set as NULL,
>>> when the QGIS attribute form loads for editing, it shows those fields as
>>> NULL inside the text field (they can be text or null). I can make changes
>>> to any fields and save and the NULL fields remain as NULL.
>>>
>>> In 3.10.2 if I go to edit the same record, the QGIS attribute form when
>>> displayed for editing the record does not show those fields as being NULL
>>> but an empty string. When I submit changes, any fields that should still be
>>> NULL are not being submitted as NULL as I get a check constraint error
>>> saying it doesn't fit the criteria. I've tried entering into those fields
>>> manually as NULL or 'NULL' or "NULL" and a dozen or so other guesses and it
>>> still fails.
>>>
>>> Again, if I roll back to version 3.10.1 it works as it has on many
>>> previous versions all the way back to 2.8
>>> _______________________________________________
>>> Qgis-user mailing list
>>> Qgis-user at lists.osgeo.org
>>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
>>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
>>
>>
>>
>> --
>> Alessandro Pasotti
>> w3:   www.itopen.it
>>
>

-- 
Alessandro Pasotti
w3:   www.itopen.it
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-user/attachments/20200204/79d7ab55/attachment.html>


More information about the Qgis-user mailing list