Emmanuele Bassi

> Usually all you need is `use gtk::subclass::prelude::*;` This is only even remotely true if all you're using is GTK. Once you start using libadwaita and gstreamer—in other words, once...

[Side note: we ended up discussing about this on Matrix] The main issue is exactly the case where the prelude ends up loading different versions of the dependencies. Yes: bad...

Something to that effect, yes. It's been more than 6 months, and I was able to figure out what the issue was. At the very least, I'd be happy with...

Ah, yes: I forgot to mention #281. Yes, this MR replaces/supersedes that one; it keeps the old introspection data, so it doesn't break existing users.

I'm happy to update the pull request to tweak some of the names, if needed. In the case of `hb_buffer_create()`, which was raised in #281, the new description is: ```xml...

> If the `_create()` methods can be made into constructors, that would be great. If not, no big deal either. I did try, but I got some mixed results. I'll...

> I am seeing these with/wihout the changes (just rebased the PR locally The `xconstructor` warnings are caused by the wrong annotation; they should be changed to `constructor`, but as...

The issue with the constructors is caused by a symbol mismatch. Every boxed type matches the `hb_gobject_` prefix, but the type falls under the `hb_` prefix. As an example, if...

Thanks for your patience. > A way to fix this would be to rename `hb_gobject__get_type()` to `hb__get_gtype()`; I guess it should be possible to maintain the old symbols in the...

> Is their anything blocking this PR? Only the CI failures.