refactor!: cleanup rules that can be assimilated as commands #220
Conversation
|
Big savings. I was hoping one day to see if the tree-sitter-nu parser would be faster than the nushell parser. However, if there's no hope for ever using it for anything other than and in-editor parser, then it's probably good to optimize it for that. |
|
@fdncred yeah, I understand the intention. However according to what I learned from the tree-sitter community, it's generally not recommended to use a tree-sitter parser in a compiler/interpreter. Besides, performance-wise, I believe the new-nu-parser(current nu-parser can be slow) should be by no means slower than TS. The only thing missing is the incremental nature. |
| (ctrl_return) @statement.outer @return.outer | ||
| (ctrl_do) @statement.outer | ||
|
|
||
| (ctrl_if) @statement.outer | ||
|
|
||
| (ctrl_try) @statement.outer | ||
|
|
||
| (ctrl_match) @statement.outer | ||
|
|
||
| (ctrl_while) @statement.outer | ||
|
|
||
| (ctrl_loop) @statement.outer | ||
| ("break" "continue") @statement.outer |
There was a problem hiding this comment.
@Bahex I've merged ctrl_do/ctrl_return/"break"/"continue" to our command rule.
I'm not sure whether some features will be missing if we remove these queries, do you have any idea?
There was a problem hiding this comment.
It should be fine as far as I can tell, and even if we want to add queries for these in the future, we can just write the them for command nodes with a specific head.
There was a problem hiding this comment.
@Bahex Would you like to review/test this PR since there are some breaking changes?
There was a problem hiding this comment.
nvim-treesitter-context has a query that would need to be updated, but other than that I haven't found anything that breaks.
Thanks for the reminder, I haven't updated the parser-rev/query-files of nvim-treesitter since cli 0.25+ was applied in this repo. |
|
Should we go ahead and land this @blindFS? |
I think so. Given there're already big gaps between the main branch and those locked by editors/plugins, it won't break things further. |
|
Thanks |
The idea of #179
It helps with the WASM compilation time #185 : case count 3146 -> 2592, time 2m20 -> 1m35
IMHO the target of tree-sitter-nu is only for code text manipulation, it's never gonna be as complete/semantically accurate as the native parser. And those rules are unnecessary in that regards.