The fact there are known workarounds for certain problems shouldn't prevent us improving and fixing the underlying causes (in a step-by-step, backwards-compatible way).
With that said, I'd rather not see Emacs go all-in on threads or GTK, since they generally feel like WorseIsBetter anti-patterns:
- GTK 2 seemed OK, but unusable for me due to the 'display disconnection' bug. GTK 3 is full of red flags: a constantly moving target, little thought given to third-party applications, pretty much killed off third-party theming, etc.
- Threading is famously error-prone, which is completely unnecessary for high-level languages like Emacs Lisp. Co-routines, async/await macros, delimited continuations; even an actor model would all seem like better fits.
GTK3 is actually feature frozen, it's not a moving target. Development has moved to GTK4. Also, it has always supported third party themes, however the convention is that they were always considered unsupported because they can break apps.
With that said, I'd rather not see Emacs go all-in on threads or GTK, since they generally feel like WorseIsBetter anti-patterns:
- GTK 2 seemed OK, but unusable for me due to the 'display disconnection' bug. GTK 3 is full of red flags: a constantly moving target, little thought given to third-party applications, pretty much killed off third-party theming, etc.
- Threading is famously error-prone, which is completely unnecessary for high-level languages like Emacs Lisp. Co-routines, async/await macros, delimited continuations; even an actor model would all seem like better fits.