The only thing in oe-core using vala is now gcr, which can optionally build vala bindings. Should vala move back to meta-oe in the 2.1 cycle?
Sounds good to me. I assume the move would include vala.bbclass? Setting documentation change to yes on that assumption ...
I'd say yes, but we'd want to put it out to the OE-Core mailing list.
Had a quick look at gcr and the vala support it asks for doesn't even get used.
But gcr needs vala's m4 files to configure. Damned m4.
Let's first resolve the gobject-introspection (which also fixes generation of vala bindings with vapigen), so that whether vala stays in oe-core or goes elsewhere, it will have those fixes.
Agreed, deferring until gobject-introspection is in.
I think we need a consultation with the community here - should Vala be dropped altogether? From what I see, the only thing that benefits from having vala and is useful in embedded environments is Rygel (the implementation of UPnP AV), and the benefit is more or less that Rygel can be patched much more easily, because it can be compiled from original vala sources instead of intermediate generated C sources.
http://lists.openembedded.org/pipermail/openembedded-core/2016-June/122632.html http://lists.openembedded.org/pipermail/openembedded-devel/2016-June/107773.html
So here's the outcome: people do appreciate having vala in oe-core, but we don't have to make sure it *really* works; as long as the recipe builds okay, and support for generating bindings doesn't bitrot, the community will handle the rest. I'll reassign the vala recipe to myself from Jussi.