Merge upstream + script to fix headers #14
Conversation
javascript: update grammar to latest version
Upgraded treesitter to latest version, v0.22.5
Updated 19 grammars to their latest versions
Update php to v0.22.4
Add SQL bindings support
Updated grammars to latest versions
Improve updater and Go -> 1.20
header_fix.py
Outdated
| 'python', | ||
| 'ruby', | ||
| 'rust', | ||
| #'typescript', Oh you are so special. You make me feel dumb |
There was a problem hiding this comment.
did you mean to commit it? i'm not sure i follow what was the intent here
There was a problem hiding this comment.
i should have left a better comment.
the typescript grammar subdir is organized slightly different from the rest of them, and there is no need to do the header fixing stuff there, and if we need to do it someday, it's very likely to be very different
| # (1) Fix up go-tree-sitter root level headers so they do not have #include "./foo.h" | ||
| # (2) Flatten a list of language subdirs that we care about. No more go-tree-sitter/{lang}/foo/bar.h | ||
| # (3) Then in these flatten subdirs, fix up any header inclusion that isn't directly a sibling file. | ||
| if __name__ == '__main__': |
There was a problem hiding this comment.
mind throwing a readme with instructions how and when this script should be used?
There was a problem hiding this comment.
added a new README_poolside.md file
| There will be lots of conflicts. Unless we have changes we specifically made for poolside's use case and have not pushed to upstream, you probably will want to prefer upstream's version in any conflict. | ||
|
|
||
| Then you need to fix up some C header #include directives, mostly because upstream is OK with having Unix shortcuts in them but Forge cannot build with such. There is a python script in this repo to help you with that: | ||
|
|
||
| ``` | ||
| # in the root of go-tree-sitter repo: | ||
| ./header_fix.py | ||
| ``` |
There was a problem hiding this comment.
Does it make sense to leverage _autoamtion/main.go for this purpose? It might be parametrised for a header prefix, or json config might be used.
There was a problem hiding this comment.
ye that's a good question.
i thought about it, but automatically generate a patch for that file is a bit harder, since it involved changing a few replaceMap into new downloadFiles. At that point we may as well just maintain a separate main.go + grammer.json fork. That's another approach
yangchi
force-pushed
the
mergeUpstream
branch
2 times, most recently
from
354edc5 to
b4ac8f8
Compare
yangchi
force-pushed
the
mergeUpstream
branch
2 times, most recently
from
af471e9 to
8ab49d4
Compare
yangchi
force-pushed
the
mergeUpstream
branch
from
8ab49d4 to
af683f7
Compare
A biggie.. i know.. You can jump directly to this commit: 85a8095
@glkz and I had some discussion about the current divergence between our fork and go-tree-sitter upstream.
The tl;dr is that getting rid of fork will be a bit hard, thanks to Bazel in Forge. But we should try to keep in sync more often.
Our fork is now behind upstream when it comes to grammar version in quite a number of languages.
This PR syncs with upstream.
The first 50 commits are from upstream
The last 2 are my own changes: The first is just merge conflicts due to the first 50.
The last commit is the only interesting one. I added a script
header_fix.pyto fix up header #include in C files. Upstream start to do the following types of inclusions that cannot work with bazelThe script tries to fix all of them by modifying these lines, and move files around.
Our smart AI wrote 90% of this python code, I did the last 10% human touch. :)
If this needs a in-person discussion then good thing I'm going to Paris to meet you all :)