cue vet playground ui

[disintegrator]

It would be nice to have a playground UI for cue vet to support users that use CUE to write validations.

[mvdan]

An alternative might be to lean into txtar files, similar to how Go's playground allows you to have multiple files: https://go.dev/play/p/BWJ4dcUqVfT

For example, one of the "Input" options could be "txtar", and then you could have something like:

cue vet foo.yaml foo.cue
-- foo.yaml --
[...]
-- foo.cue --
[...]

Then the output would be that of the command at the top. As a first rough pass, we could only allow one command at the top.

The reason I bring this up is because it's more flexible - it supports arbitrary input and output formats, any command that cmd/cue supports, etc. Then we don't need to keep expanding the dropdowns to support every possible combination a user might want to do.

I see that @myitcv already mentions txtar in a TODO in https://github.com/cue-lang/cuelang.org/tree/master/play; I'm not sure if this is what he had in mind.

[disintegrator]

Oh TIL. I've seen @myitcv use this format and I would personally be happy to use it. However, it's going to feel foreign to non-Go users and perhaps the UI can be a cover over the txtar representation...

[myitcv]

However, it's going to feel foreign to non-Go users and perhaps the UI can be a cover over the txtar representation...

100% agree. For some people, they will prefer the light touch of txtar. For others it will be an implementation detail that can be hidden using good UI/UX.

As you will have noticed from the playground, I score 0/10 on UI/UX. It was a weekend hack project early in the day when I was exploring CUE. Delighted to have others opine on how it can be improved.

[verdverm]

A devcontainer could be far less work with way more end user functionality

• https://docs.github.com/en/codespaces/setting-up-your-project-for-codespaces/introduction-to-dev-containers
• https://github.com/dagger/dagger/tree/main/.devcontainer

[myitcv]

Definitely something we should look at. Indeed, as @disintegrator mentioned on Slack, I don't think investigating this route is mutually exclusive with improving the playground.