luvit
Update get-list scripts to support lit 3.9.0 and luvi 2.15.0
Some consideration should be applied that this change will cause anyone using get-lit with the $LUVI_VERSION environment variable less than 2.15.0 (such as potentially in automated testing or likewise) to suddenly start failing.
Sounds good, except I'm too lazy to write logic for semver comparisons in languages that I don't know. I'll change this to a draft for now.
A pure sh semver comparison is going to be annoying. The easiest way is to depend on gnu coreutils such as sort -V. should be fine.
Alternatively, the script can fallback to the old url format if the new one fails.
Honestly this change is also a pretty good chance for us to rework how this works in the first place, what are thoughts on creating a new repository that packages releases for luvi+lit+luvit all in one .tar.xz or .zip.
Then git-lit can be simplified to a single download + extract and we completely bypass handling any weird edge cases like how to handle one download failing and one succeeding.
Sounds good, but do we need a new repo for that? I feel like a number of existing repos can already fulfill that. Would also be neat if the action automatically triggers on releases/tag creation
My only hesitation to providing prebuilt luvit and lit binaries is that this opposes our message that luvi apps are supposed to be easy to build on the fly. It is convenient, though, so I support it overall. I do not have an opinion on how they are hosted.
I've been side tracked doing other things so I don't think I'll get around to finishing up an improved get-lit script anytime soon. I say it's probably reasonable to pull the trigger on this and merge it and deal with any side effects if they pop up.
If someone really wants to use the old get-lit there's nothing preventing them from using an older version of the script using a specific commit instead of master.
This isn't a perfect solution, but it is a solution and introduces the newest luvi and lit into the ecosystem.
I think you should re-open this and just merge it, as far as I am aware there aren't really any automated tests/actions manually specifying the versioning, and if any as truemedian said we can handle it per case, it's not really worth the extra effort.
Cannot re-open this one directly, need to write a new PR. Because of our Thanksgiving holiday, remind me again after the weekend, unless someone else does this first.