#CUE SDK
1 messages · Page 1 of 1 (latest)
I don't see it as being deprecated. We're making an effort to avoid breaking past configs from Europa (v0.2).
It would be easier to skip to light bindings (I think).
A lot of people wanted to use other languages other than CUE (like Go and Typescript) but we don't have a good idea on how many want to keep it even though we know there are at least a few. It does allow for a declarative pipeline that is better than YAML for those who prefer it.
So it's this uncertainty of interest that makes us say "we'll see".
Right, I kind of like being able to use a declarative config language to describe stuff without having to cross into a full-blown programming language territory, so I would kind of miss it, if it ever went away. Which is why I'm asking about it.
But I can also appreciate how different CUE is from the norm (probably even more alien than Nix or the like) and that even if you're used to weird languages it's still a bit rough around the edges (I remember having syntax errors presenting as unification errors or having to emulate functions with structs as being annoying) – so I can see how having a priori commitment to it at this time might not make sense.
That said, I would appreciate if you considered having some kind of a not-a-programming-language SDK in the future (if not CUE, there's always other options like Jsonnet or Nickel or whatever).
We are open to any SDK that has any interest from the community. You can vote here: https://github.com/dagger/dagger/discussions/3489.
CUE isn't up for voting because we're already supporting it. The question remains if it'll continue with the same API or will it evolve as a light binding for the 0.3 API. To drop it entirely would mean that no one wants it (apart from a handful), and I don't think that'll happen.
It also depends on contribution. We're just prioritizing based on resources, but anyone can contribute an SDK assuming someone can also maintain it.