Why does Layout::set_text need a length argument? about pango HOT 15 CLOSED

https://bugzilla.gnome.org/show_bug.cgi?id=784021
IMHO we should just patch the Pango .gir file instead of adding configuration. Configuration only makes sense for things that can't be expressed by the .gir file.

from pango.

Then let's close it. :)

from pango.

It's part of the things we need to improve. However I was certain that we removed the next argument if it was an integer and the current one was a string. Did I miss something @EPashkin?

from pango.

I can't see any code in GIR that would do that, but I might be looking for the wrong thing.

I can see that CssProvider.load_from_data() is implemented manually in order to hide its length parameter. Would it be appropriate to do something similar here, or is a generic solution in GIR preferable? I'd be happy to have a go at fixing this in either case.

from pango.

It still under construction gtk-rs/gir#376
I think we accept manual version until issue fixed.

from pango.

I see gtk-rs/gir#376 is in progress, so I think it's fine to wait for it to be done automatically.

from pango.

This is not covered unfortunately because the array-length annotation is missing in Pango. Someone should file a bug in Bugzilla to get this added, or I'll do it if nobody else is faster.

from pango.

Maybe we need add config override for array-length?

from pango.

That would make sense (and is relatively simple), yes. Nonetheless all these things should be reported to the respective C libraries.

from pango.

Currently we strong against patching .gir files as it will be replaced by next update.
Or we need patching mechanism that works good with xml files.

from pango.

Next upstream release would ideally have these changes so no patching would be needed after an update.

from pango.

Maybe but I don't want look change in big xml and think, fixed it actually or not :(

from pango.

The changes between versions are not actually very big usually, unless you skip many major versions. When doing GStreamer releases, I usually go through the diff of the .gir files and it was a manageable task so far. Usually there are not many changes/additions to already existing types.

Problem I see with a configuration is that it "makes lazy", and then the annotations are never fixed in the C libraries and every binding ships its own configuration around for these bugs.

from pango.

See gtk-rs/gir#387 (comment) . Needs manual bindings for these functions that take a string+length, or return a string+length

from pango.

🎉 It looks like this is fixed in the latest pango release: LayoutExt::set_text(&str)

from pango.