script: Prescan byte stream to determine encoding before parsing document#41376
Merged
simonwuelker merged 9 commits intoservo:mainfrom Dec 19, 2025
Merged
script: Prescan byte stream to determine encoding before parsing document#41376simonwuelker merged 9 commits intoservo:mainfrom
simonwuelker merged 9 commits intoservo:mainfrom
Conversation
servo-highfive
added
the
S-awaiting-review
simonwuelker
commented
Dec 18, 2025
Comment on lines
+1
to
2
| [in-script.html] | ||
| expected: FAIL |
Contributor
Author
There was a problem hiding this comment.
These charset/in-* tests fail because the prescanning algorithm doesn't know that it should ignore things like <script><meta charset="windows-1251"></script>. I'll fix this in a followup, since this PR is large enough already.
simonwuelker
force-pushed
the
sniff-encodings
branch
from
7047f44 to
2766716
Compare
mrobinson
approved these changes
Dec 18, 2025
Member
mrobinson
left a comment
mrobinson
left a comment
There was a problem hiding this comment.
Amazing work @simonwuelker!
Comment on lines
+164
to
+177
| if let Some(encoding) = encoding_detector.buffer(chunk) { | ||
| document.set_encoding(encoding); | ||
| let buffered_bytes = mem::take(&mut encoding_detector.buffered_bytes); | ||
| *self = Self::Decoding(NetworkDecoder { | ||
| decoder: Some(LossyDecoder::new_encoding_rs( | ||
| encoding, | ||
| NetworkSink::default(), | ||
| )), | ||
| encoding, | ||
| }); | ||
| return self.push(&buffered_bytes, document); | ||
| } | ||
|
|
||
| None |
Member
There was a problem hiding this comment.
Suggested change
| if let Some(encoding) = encoding_detector.buffer(chunk) { | |
| document.set_encoding(encoding); | |
| let buffered_bytes = mem::take(&mut encoding_detector.buffered_bytes); | |
| *self = Self::Decoding(NetworkDecoder { | |
| decoder: Some(LossyDecoder::new_encoding_rs( | |
| encoding, | |
| NetworkSink::default(), | |
| )), | |
| encoding, | |
| }); | |
| return self.push(&buffered_bytes, document); | |
| } | |
| None | |
| encoding_detector.buffer(chunk).map(|encoding| { | |
| document.set_encoding(encoding); | |
| let buffered_bytes = mem::take(&mut encoding_detector.buffered_bytes); | |
| *self = Self::Decoding(NetworkDecoder { | |
| decoder: Some(LossyDecoder::new_encoding_rs( | |
| encoding, | |
| NetworkSink::default(), | |
| )), | |
| encoding, | |
| }); | |
| return self.push(&buffered_bytes, document); | |
| }) |
Contributor
Author
There was a problem hiding this comment.
Assigning to *self in the closure doesn't seem to be accepted by the borrow checker, because it also needs encoding_detector.buffered_bytes, which is borrowed from self.
Comment on lines
+2
to
+4
| expected: ERROR | ||
| [WebGL test #3] | ||
| expected: FAIL |
Contributor
Author
There was a problem hiding this comment.
Likely a unrelated intermittent issue that I missed in my try run (:
servo-highfive
removed
the
S-awaiting-review
simonwuelker
force-pushed
the
sniff-encodings
branch
from
2766716 to
586a9f2
Compare
servo-highfive
added
the
S-awaiting-review
simonwuelker
force-pushed
the
sniff-encodings
branch
from
586a9f2 to
768d35a
Compare
|
🔨 Triggering try run (#20350147469) for Linux (WPT) |
simonwuelker
force-pushed
the
sniff-encodings
branch
from
768d35a to
eabd62b
Compare
|
Test results for linux-wpt from try job (#20350147469): Flaky unexpected result (24)
Stable unexpected results that are known to be intermittent (31)
|
|
✨ Try run (#20350147469) succeeded. |
servo-highfive
added
the
S-awaiting-merge
github-merge-queue bot
pushed a commit
that referenced
this pull request
Dec 18, 2025
…ment (#41376) Servo currently completely ignores `<meta charset>` tags. When we find one with an encoding that is incompatible to the current one, then we should reload the page and start over with the new encoding. A common optimization that has even made its way into the specification is to wait for a few bytes to arrive and inspect them for `meta` tags, so the browser is able to use the correct encoding from the very beginng. In practice, I've run into problems with our WPT harness when reloading the page after `meta` tags. Therefore, this change implement the optimization first, so we never have to reload when running WPT. I've implemented prescanning in a way where we wait for 1024 bytes to arrive or for one second to pass, whichever one happens first. This causes a large number of web platform tests to flip around. I've looked at most of the new failures and I believe they're reasonable. Testing: New tests start to pass. Part of #6414 --------- Signed-off-by: Simon Wülker <[email protected]>
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
servo-highfive
added
S-tests-failed
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
Signed-off-by: Simon Wülker <[email protected]>
simonwuelker
force-pushed
the
sniff-encodings
branch
from
eabd62b to
ea7333d
Compare
servo-highfive
removed
the
S-tests-failed
Contributor
Author
|
Found a source of intermittency (22403fc) and fixing it gives us another 300 passes :) |
|
🔨 Triggering try run (#20364703193) for Linux (WPT) |
|
Test results for linux-wpt from try job (#20364703193): Flaky unexpected result (22)
Stable unexpected results that are known to be intermittent (33)
Stable unexpected results (17)
|
|
|
Signed-off-by: Simon Wülker <[email protected]>
simonwuelker
force-pushed
the
sniff-encodings
branch
from
ea7333d to
004ab4f
Compare
servo-highfive
added
the
S-awaiting-merge
servo-highfive
removed
the
S-awaiting-merge
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Servo currently completely ignores
<meta charset>tags. When we find one with an encoding that is incompatible to the current one, then we should reload the page and start over with the new encoding. A common optimization that has even made its way into the specification is to wait for a few bytes to arrive and inspect them formetatags, so the browser is able to use the correct encoding from the very beginng.In practice, I've run into problems with our WPT harness when reloading the page after
metatags. Therefore, this change implement the optimization first, so we never have to reload when running WPT. I've implemented prescanning in a way where we wait for 1024 bytes to arrive or for one second to pass, whichever one happens first.This causes a large number of web platform tests to flip around. I've looked at most of the new failures and I believe they're reasonable.
Testing: New tests start to pass.
Part of #6414