QGIS » QGIS Application

several users on windows working on the same database in an office LAN environment have reported the same issue. I just tested and confirmed it on Linux over the Internet.

Giovanni Manghi wrote:

Gavin Fleming wrote:

no I tested as superuser on the same layer filtered to one record. The only difference was the QGIS version.

as I cannot replicate in any way i leave this open so others would be able to give it a try and leave feedback.

PS
If it affects master and it affects editing of PostGIS layers please leave the proper tags.

I was able to replicate here against Gavin's db. Under QGIS 2.6 the following simple expression fails to update existing field:

"type" + ')'

Under 2.4 this works:

"type" || ')'

I just tested using the python API:

l = iface.activeLayer()
exp = QgsExpression('"type" + \\')\\'')
f = l.getFeatures()
for feat in f:
    value = exp.evaluate(feat)
    print value

And that works fine, so the issue must lie in the caclulator itself.

I was able to replicate here against Gavin's db. Under QGIS 2.6 the following simple expression fails to update existing field:

[...]

Under 2.4 this works:

[...]

so it is just this particular expression (concatenation with the "+") that fails or any expression fails?

and then: it fails on any database or just one? and what about other formats?

I tested here on local and remote databases, with qgis 2.6 and master and both ways to do concatenation (|| and +) are working as expected from the field calculator.

the concatenation is a red herring :-) - any expression fails. Remote vs local might be a red herring too.
2.4 works on shapefiles, postgresql 9.1 and postgresql 9.3
2.6 works on shpefiles, postgresql 9.3 but NOT postgresql 9.1

It does seem to be the same or related issue. I tested this one again and noticed the extra field that gets created as mentioned in #11538, which I hadn't noticed before.

I have the same issue. When I open the table of attributes, choose to just show the selected items and modify a field value using the toolbar of the table of attributes dialog, it works fine. But when I open the field calculator dialog and try to modify some values, it didn't apply the modifications.

Tim Sutton wrote:

I just tested using the python API:

[...]

And that works fine, so the issue must lie in the caclulator itself.

Wille Marcel wrote:

I have the same issue. When I open the table of attributes, choose to just show the selected items and modify a field value using the toolbar of the table of attributes dialog, it works fine. But when I open the field calculator dialog and try to modify some values, it didn't apply the modifications.

always? or just in same situation as #11538?

because for normally imported vectors then is works as expected...

Only with postgis. When editing vector files it works properly.

Giovanni Manghi wrote:

Wille Marcel wrote:

I have the same issue. When I open the table of attributes, choose to just show the selected items and modify a field value using the toolbar of the table of attributes dialog, it works fine. But when I open the field calculator dialog and try to modify some values, it didn't apply the modifications.

always? or just in same situation as #11538?

because for normally imported vectors then is works as expected...

Wille Marcel wrote:

Only with postgis.

yes, but the ticket I linked to you is about (and this one too) is about postgis tables (layers) created in a specific way. If you import a vector (shape, whatever) into postgis using qgis (db manager, browser. etc), shp2pgsql, ogr2ogr, etc. then this does not happens, or at least I can't confirm in any test I made so far.

I tried it in a OSM database created with osm2pgsql and in a database with all the vectors created using qgis. In both the error occurs.

Giovanni Manghi wrote:

Wille Marcel wrote:

Only with postgis.

yes, but the ticket I linked to you is about (and this one too) is about postgis tables (layers) created in a specific way. If you import a vector (shape, whatever) into postgis using qgis (db manager, browser. etc), shp2pgsql, ogr2ogr, etc. then this does not happens, or at least I can't confirm in any test I made so far.

Wille Marcel wrote:

I tried it in a OSM database created with osm2pgsql and in a database with all the vectors created using qgis. In both the error occurs.

can you attach a dump of one of the table that has the issue?

appears to be fixed in master. The value updates in the table, detects that there are changes and saves, i.e. as expected. Database and PostGIS version and test data are the same as those that had this issue with 2.6 and earlier versions of master.