[Qgis-psc] Attachment widget broken in the attribute table

Matthias Kuhn matthias at opengis.ch
Tue Aug 6 08:02:05 PDT 2019


Hi PSC,

As discussed with Giovanni Manghi in [1],  I'd like to bring an issue 
with the attachment widget to your attention.

Since QGIS 3.0 there has been an issue with the attachment widget, that 
when clicking the [...] browse button in the attribute table, QGIS has 
crashed. This has been reported in [2] about 1.5 years ago.

In the last months, at least 3 core developers have spent a total of at 
least 10 hours trying to recover the original browse functionality. 
Unfortunately without success.

Finally, a pull request was created to remove the button from the 
attribute table [1] and has been merged into master (backports for 3.4 
and 3.8 are pending).

There was a broad agreement that at the current point we are not able to 
fix it in a reasonable timeframe, so it's better to remove the button 
(which had one sole functionality of crashing QGIS). All other 
functionality of the attachment widget (outside the attribute table) is 
unaffected.


One of the conclusions in [1] was, that it would be good to collaborate 
with Qt (e.g. KDAB) for bugfixing. Apart from this particular issue, 
there are a few others for which an upstream fix is the only (or the 
cleanest) solution. It would be good if we could agree on a budget for 
this (e.g. 5 days of time) and get in touch with upstream to get the 
ball rolling. It would be nice if the PSC could take a decision and do 
the preparation work for that (if possible pre 3.10).

On another note, it would be good to clarify on our stance regarding 
priority handling. In the discussion there was a pretty tight notion of 
a "no regressions allowed, whatever the cost" policy. This binary way 
this argument was used made it quite hard to decide on the best way 
forward for the application. I have the plan to try to avoid regressions 
wherever possible and to make QGIS more stable at the same time. But if 
the project decision is that stability shouldn't be much of a focus 
before anything tagged a "regression" is eliminated I will have to check 
my priorities again.

Thanks a lot

Matthias

[1] https://github.com/qgis/QGIS/pull/30930#issuecomment-518660334

[2] https://github.com/qgis/QGIS/issues/26948




More information about the Qgis-psc mailing list