feat: pipe operators

[mightyiam]

Closes #83

[theutz]

I'm curious if there's any reason this hasn't been merged yet? Also curious if anyone's had any success using this fork in NotAShelf/nvf? I couldn't figure out how to make it work.

[mightyiam]

No one prioritized it. I think this is just a quick review.

[adisbladis]

This is not trying to block a merge, but I think it's an important policy/philosophical question to answer:
Should a (semi)-official grammar support experimental features? I don't think so (merging experimental features into "core" tooling is what lead to Flakes being so controversial).
It's a potential backdoor into stabilising features that we don't really have a consensus on yet.

[adisbladis]

Re the pipe operator itself (irrelevant to the overall merging or not of this PR):
I personally heavily dislike the pipe operator and would like to see the feature reverted.

[blindFS]

This is not trying to block a merge, but I think it's an important policy/philosophical question to answer:

Should a (semi)-official grammar support experimental features? I don't think so (merging experimental features into "core" tooling is what lead to Flakes being so controversial).

It's a potential backdoor into stabilising features that we don't really have a consensus on yet.

Just a few more operators won't hurt anything. If the feature gets aborted upstream, it's also pretty easy to revert this.

[khaneliman]

This is not trying to block a merge, but I think it's an important policy/philosophical question to answer:

Should a (semi)-official grammar support experimental features? I don't think so (merging experimental features into "core" tooling is what lead to Flakes being so controversial).

It's a potential backdoor into stabilising features that we don't really have a consensus on yet.

It's irrelevant if it's experimental or not. A parser/query/lsp is a reactive optional tooling that needs to know how the language works now. Its inclusion should have no effect on the stabilization. If nix reverts the feature there is no drastic side effect to treesitter parser finding significance in 2 characters being next to each other.

The flip side opinion would be how you can you properly evaluate an experimental feature if you never update tooling to work with it.

[theutz]

This is not trying to block a merge, but I think it's an important policy/philosophical question to answer: Should a (semi)-official grammar support experimental features? I don't think so (merging experimental features into "core" tooling is what lead to Flakes being so controversial). It's a potential backdoor into stabilising features that we don't really have a consensus on yet.

@adisbladis I would love to discuss an issue like that, and honestly I have very little stake in the outcome here. I've just been relatively annoyed by the errors I get in Neovim when I use the pipe operators. :)

That being said, since starting a discussion about the merits of an experimental feature here might drown the work that @mightyiam did to make this PR, is there a place where I (we?) can read more about your thoughts on pipe operators and what you dislike about them? I find that understanding the opinions of more experienced developers with different perspectives than my own is one of the quickest ways for me to learn.