[2.12] Filtering feature stopped working

[hartwork]

Hi @tibirna,

we received a bug report for QGit 2.12 from @joehni downstream in Gentoo. He wrote:

QGit contains a filter mechanism in the toolbar where the user can search for a given string in a selected area of the commit. The area is selected using a drop down box, the filter is applied when the button with the filter symbol right of the edit field is toggled. However, this no longer works. As soon as a search text is entered and the filter is applied, not a single commit matches anymore and the list of commits is empty no matter what search area is selected.

From a first quick try myself, my impression is that:

• QGit 2.12 does indeed show this buggy behavior.
• QGit 2.11 does not show this behavior.

What do you think?

Best, Sebastian

[tibirna]

Yes. This was reported with #162 and fixed with 757c0fc. I didn't make a new release yet because there are some additional issues with opening the external diff viewer, probably also caused by the migration to Qt6 (and the RegExp "windstorm"). I was hoping to fix that too, and perhaps have a contributed solution for the breakage of the Qt4 support, before issuing a new release. I'm aware of all this, I'll address it in the near future.

Thanks.

[hartwork]

Yes. This was reported with #162 and fixed with 757c0fc.

@tibirna I see, thanks! Should we close #165 here as a duplicate then?

I didn't make a new release yet because there are some additional issues with opening the external diff viewer, probably also caused by the migration to Qt6 (and the RegExp "windstorm").

Sounds like #163.

I was hoping to fix that too, and perhaps have a contributed solution for the breakage of the Qt4 support, before issuing a new release.

Sounds like #161. Personally I consider Qt4 support a mistake in use of time and additional complexity. PowerPC-supporting OS X 10.6 is dead and QGit will not change that. I lterally have someone sitting next to me with a PowerPC Apple somewhere that they "would not put on the Internet anymore today" because it is unsupported for more than 14 years by now.

[tibirna]

Yes, I will close this (#165) when I will close #162 too.

Support for compiling with Qt4 is not a mistake as long as there is a developer willingly investing their time and effort to maintain it. This is not the blocking issue for the next release, though. The external diff viewer issue is. When I will have that fixed, I will ponder releasing 2.13.

What nags at me more is that I released 2.12 with insufficient testing and thus with quite unfortunate regressions. This I need to handle with more care in the future.

[hartwork]

Support for compiling with Qt4 is not a mistake as long as there is a developer willingly investing their time and effort to maintain it.

I think that picture misses the increase in complexity (and its cost on others) and the time the review takes from others. You seem to be willing to carry that cost yourself (or have fun at it), but it's not free.