Kmeakin edited PR #9108.
Kmeakin edited PR #9108:
- Instead of pushing every byte in a lexed integer into a
Vec<u8>
, use a slice of the original buffer, and only reallocate if there are underscores that need to be removed- Don't track line/column in
Pos
, so don't need to update line/column inadvance_pos()
Kmeakin edited PR #9108:
- Instead of pushing every byte in a lexed integer into a
Vec<u8>
, use a slice of the original buffer, and only reallocate if there are underscores that need to be removed- Don't track line/column in
Pos
, so don't need to update line/column inadvance_pos()
- Store text being lexed as
&str
rather than&[u8]
to avoid checking substrings for UTF-8 validty
Kmeakin updated PR #9108.
Kmeakin updated PR #9108.
Kmeakin updated PR #9108.
Kmeakin updated PR #9108.
Kmeakin updated PR #9108.
Kmeakin edited PR #9108:
- Instead of pushing every byte in a lexed integer into a
Vec<u8>
, use a slice of the original buffer, and only reallocate if there are underscores that need to be removed- Don't track line/column in
Pos
, so don't need to update line/column inadvance_pos()
- Store text being lexed as
&str
rather than&[u8]
to avoid checking substrings for UTF-8 validity- Don't try to parse integers twice
github-actions[bot] commented on PR #9108:
Subscribe to Label Action
cc @cfallin, @fitzgen
<details>
This issue or pull request has been labeled: "cranelift", "isle"Thus the following users have been cc'd because of the following labels:
- cfallin: isle
- fitzgen: isle
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
pchickey commented on PR #9108:
Hi, thanks for these contributions! If I can nerd-snipe you into some unrelated lexer work, we have a
logos
lexer over in thewasm-wave
crate that we need to be rewritten entirely in safe rust. If you are interested in learning more, ping me over on https://github.com/bytecodealliance/wasmtime/pull/8872
fitzgen submitted PR review:
This is fantastic, thanks!
Apologies for the slow review turn around, I've been having ISP issues.
One comment below that I'd like to get your take on, but shouldn't be a blocker for landing this. Can address it in a follow up, if necessary.
fitzgen submitted PR review:
This is fantastic, thanks!
Apologies for the slow review turn around, I've been having ISP issues.
One comment below that I'd like to get your take on, but shouldn't be a blocker for landing this. Can address it in a follow up, if necessary.
fitzgen created PR review comment:
Why not
fn peek_char(&self) -> Option<char>
? I guess a byte is mildly more performant, in theory if LLVM doesn't clean things up, but that doesn't seem super compelling compared to the simplification of "we are dealing with strings, so we should also deal with chars".
Kmeakin updated PR #9108.
fitzgen merged PR #9108.
Last updated: Dec 23 2024 at 12:05 UTC