The last 3 times I've worked on Vala code for Builder, I've ran into something broken in the compiler with regards to G-I integration.
Like, parameter order scrambling type issues. In otherwords, longjump's into data...
and it's so damn frustrating, because I constantly have to dumb down my APIs to get things to work there.
I almost never have this problem with PyGObject.
@hergertme I definitely think of Vala as C shorthand and end up consulting generated code frequently. the semantics aren't well-documented and the static checks are often insufficient to actually be confident your code is correct without reading the C...
imo #rustlang is what Vala wanted to be, plus it's faster and uses less RAM...
@martensitingale I prefer to write c, and hope to retire before that has to change.
@hergertme I don't really have any solutions to offer but I can commiserate with you about Vala - even from the perspective of someone just trying to use it for one of its most basic use-cases (writing a GUI desktop app for GNOME), it's kind of challenging to figure out the Rube Goldberg machine to make it actually WORK.
@clee Solutions are rather easy. I file bugs, they usually get fixed.
The number of corner cases in a compiler are huge. Vala trying to have multiple layers (vapi, g-i, code) does not make their problem space smaller.
@hergertme This is hardly surprising given that despite being both volunteer run projects, vala has not only fewer contributors but also a much larger scope compared to the python bindings.
I know it's kind of cool to do so, and I understand it is frustrating, but ragging on vala lime this is pretty unhelpful. It's a useful tool in the ecosystem at the moment.
@mjog Spare me. I deal with this all the time and I always hold my tongue because of how vocal there community is to backlash against criticism.
Nothing I said was ragging, it was pure description of what I deal with on a regular basis.
@hergertme Hey, as someone who inherited a non-trivial vala project, I know full well it's quirkiness.
But complaining about it being buggier than a smaller, much more mature project is still unfair. Especially given the work that has been put in to vala over the last 12 months.
Constructive criticism is useful, but "vala sucks" (which most/all of the recent criticism has been) isn't. File bugs, attach patches. You know the drill.
@mjog Obviously bugs were filed. But maybe instead of writing a new completion engine for Vala, I'll just delete the code and go work on something else that doesn't involve self-righteous people trying to make me feel bad for being realistic about my experiences on my personal mastadon account.
@hergertme Hey yo, I'm sorry for making you feel like that. I'll shut up then.
@mjog Apology accepted, water under the bridge.
In many ways, I'd be better off implementing the Vala support in Builder in C and calling the generated C API for libvala because at least I'd know my code is correct.