Open champignoom opened 7 months ago
I think switching the default branch is a bad idea but adding an entry about 7.0.0 sounds good! I'll do that soon
@vhyrro I just made a few changes, since v7.0.0 itself has some highlighting issues with modern treesitter and kanagawa.
I'm actually against this. External modules will need to test to work with two versions of neorg which is pretty difficult sometimes.
well if we've somewhen fixed the luarocks things there hopefully should never be so many big issues again
well if we've somewhen fixed the luarocks things there hopefully should never be so many big issues again
Maybe, but the main point is about a notion of "stable branch", which would still help when (or especially when) a good package manager is ready.
The new actions should boil down to:
somehow constantly backporting bug fixes into the stable branch have fun with git constantly picking commits from one branch to another
(instead of all the "v?.?.?" tags) that's the whole point of semver which we decided to use
@champignoom
From what I observe, people that cannot install luarocks.nvim are either having skill issues :kek: or are on termux.
The one you linked above is an issue with rocks.nvim and nvim-treesitter not being able to package a rock correctly. This has nothing to do with neorg afaik. Explained very well in https://github.com/nvim-neorg/neorg/issues/1351#issuecomment-2021708858
having skill issues
Actually in 99% it's skill issue or just doing like 3 things with e.g. Lazy build xyz
and things are fixed
One thing that isn't a skill issue though is lazy sometimes not running the build script, which I've tried tracking down in the code but the code seems to be behaving just fine...
If we can track down that one problem then we can guarantee that luarocks installation will be correct under 95% of cases (with the remaining 5% being windows users :p). I like the notion that's presented here, but I think it's not the correct call.
Users are enouraged to stay on 7.0.0 if they don't feel like 8.0.0 is giving them the stable experience. The stable
tag would only point to 7.0.0
at best, after which it becomes frankly useless once luarocks support is properly stabilized, leaving me to manually update the stable release every time release-please issues a new release tag. So many plugins that initially tried to have a stable branch decided to screw it and move to a release model like ours instead, including neorg several years ago when we had both a dev
and a stable
branch. I'll add the mention to the README about 7.0.0 right now :)
From what I observe, people that cannot install luarocks.nvim are either having skill issues :kek: or are on termux.
The one you linked above is an issue with rocks.nvim and nvim-treesitter not being able to package a rock correctly. This has nothing to do with neorg afaik.
@pysan3 That's an issue within dependencies that neorg chooses to depend on. Leaving it to possibly non-dev users to solve is a sign of leaky abstraction. Neorg could have been able to just work, with no excuses.
The
stable
tag would only point to7.0.0
at best, after which it becomes frankly useless once luarocks support is properly stabilized, leaving me to manually update the stable release every time release-please issues a new release tag.
@vhyrro I agree that luarocks would potentially make a fundamental difference and solve the mess of versioning altogether. But until it's stablized, or until a tutorial of stably working with an unstable luarock is ready, tags like 8.x.y-alpha
could be more informative.
Issues
Feature description
The whole luarocks system since v8.0 seems to be not stable enough yet, see for example https://github.com/nvim-neorg/neorg/issues/1351.
A kickstart config that takes care of all the glitches also seems to be still on the way, see https://github.com/nvim-neorg/neorg/issues/1366.
Until then, I suggest to add a more visible and more fool-proof
stable
branch. For example:stable
branch from a pre-v8 working commit, like https://github.com/nvim-neorg/neorg/commit/086891d396ac9fccd91faf1520f563b6eb9eb942, and optionally make it default;commit = "086891d396ac9fccd91faf1520f563b6eb9eb942",
in the TLDR section of stable/READMEHelp
No
Implementation help
No response