Closed
Bug 1324193
Opened 8 years ago
Closed 8 years ago
Bump rust-url to 1.2.4
Categories
(Core :: Networking, defect)
Core
Networking
Tracking
()
RESOLVED
FIXED
mozilla53
Tracking | Status | |
---|---|---|
firefox53 | --- | fixed |
People
(Reporter: manishearth, Assigned: manishearth)
References
Details
Attachments
(1 file)
Brings in some crash fixes and correctness updates. Any further mismatches *should* be due to things needing changes in nsStandardURL, not rust-url.
Comment hidden (mozreview-request) |
Assignee | ||
Comment 2•8 years ago
|
||
Try push with rust-url enabled and logs turned into printfs at https://treeherder.mozilla.org/#/jobs?repo=try&revision=03032638e05f69570849171753f04eebfe318b4b
Assignee | ||
Comment 3•8 years ago
|
||
Ah, forgot to turn off the fallback mode.
Assignee | ||
Comment 4•8 years ago
|
||
Added talos to see if the perf impact at startup is measurable. If not, we'll have to manually try to compare release builds.
New try: https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca
Perfherder: https://treeherder.mozilla.org/perf.html#/compare?originalProject=mozilla-central&originalRevision=34a1ab064cb5&newProject=try&newRevision=9b09455e1f766fc9f215057f0a1012c3119892ca&framework=1&showOnlyImportant=0
Assignee | ||
Comment 5•8 years ago
|
||
Now the failures are:
IDNA issues all over the place (bug 945240)
> TEST-UNEXPECTED-FAIL | dom/tests/mochitest/whatwg/test_postMessage_idn.xhtml | wrong origin -- IDN issue, perhaps? - got "http://sub1.xn--lt-uia.example.org:8000", expected "http://sub1.ält.example.org:8000"
IPv4-in-ipv6 being printed as hex (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931324)
> TEST-UNEXPECTED-FAIL | dom/url/tests/test_url.html | IPv6 hostname - got "[::c009:505]", expected "[::192.9.5.5]"
> TEST-UNEXPECTED-FAIL | dom/url/tests/test_url.html | undefined assertion name - got "http://[::c009:505]/", expected "http://[::192.9.5.5]/"
c: drive stuff -- this looks like a gecko bug? (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931291)
> TEST-UNEXPECTED-FAIL | layout/style/test/test_bug379440.html | Serialize unloadable URLs using their specified value - got "url(\"file:///tmp/foo\"), url(\"file:///c:/\"), url(\"http://example.com/\"), crosshair", expected "url(\"file:///tmp/foo\"), url(\"file:///c|/\"), url(\"http://example.com/\"), crosshair"
Tons of UNEXPECTED-PASSes on WPT https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931347
> TEST-UNEXPECTED-PASS | /url/a-element-xhtml.xhtml | Parsing: <http://foo.com:b@d/> against <http://example.org/foo/bar> - expected FAIL
ipv6 double colon (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931412)
> TEST-UNEXPECTED-FAIL | docshell/test/unit/test_nsDefaultURIFixup_info.js | do_single_test_run - [do_single_test_run : 584] "http://[fe80:cd00::cde:1257:0:211e:729c]/" == "http://[fe80:cd00:0:cde:1257:0:211e:729c]/"
Empty ftp (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931563)
> TEST-UNEXPECTED-FAIL | netwerk/test/unit/test_URIs.js | do_check_property - [do_check_property : 319] "" == "ftp://"
Differing percent encoding (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931563)
> TEST-UNEXPECTED-FAIL | netwerk/test/unit/test_bug376844.js | run_test - [run_test : 19] "http://example.com/?'" == "http://example.com/?%27"
> TEST-UNEXPECTED-FAIL | netwerk/test/unit/test_standardurl.js | test_filterWhitespace - [test_filterWhitespace : 314] "http://test.com/pa%0D%0A%09th?query#hash" == "http://test.com/pa%0D%0A%09th?qu%0D%0A%09ery#hash"
Something else? (https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b09455e1f766fc9f215057f0a1012c3119892ca&selectedJob=32931602)
> TEST-UNEXPECTED-FAIL | security/manager/ssl/tests/unit/test_sts_fqdn.js | run_test - [run_test : 46] "" == ".."
Not sure which of these are rust-url bugs.
Assignee | ||
Comment 6•8 years ago
|
||
> IDNA issues all over the place
bug 945240
> IPv4-in-ipv6 being printed as hex
https://url.spec.whatwg.org/#concept-ipv6-serializer
Chrome agrees with the spec
bug 1324243
> c: drive stuff
bug 1324251
Chrome, Firefox, Safari agree here.
Tests disagree: https://github.com/w3c/web-platform-tests/blob/be5820ecd36650bff4728208629553931fd42ec2/url/urltestdata.json#L1337
These are webkit's own tests (https://trac.webkit.org/browser/trunk/LayoutTests/fast/url/file.html), so they seem to disagree with themselves? Webkit trunk might pass these though.
https://url.spec.whatwg.org/#windows-drive-letter
> Tons of UNEXPECTED-PASSes on WPT
yay. Not sure if I should file bugs for fixing these in nsStandardURL? It seems to be a whole bunch of bugs.
> ipv6 double colon
Not sure how fixups will interact with rust-url but this specific bug is due to https://github.com/servo/rust-url/issues/253
> Empty ftp
bug 1324244
> Differing percent encoding
discussion at https://github.com/whatwg/url/issues/172
gecko bugs: bug 1324246, bug 1324247
(consistent in firefox and chrome, does not match spec or webkit.)
> Something else?
https://github.com/servo/rust-url/issues/255
Assignee | ||
Comment 7•8 years ago
|
||
Perfheder reports 4-8% increase on most of the tests I ran. To be expected.
(I wanted to run perfherder to get an idea of the baseline perf impact when we start attempting to reduce it)
Comment 8•8 years ago
|
||
mozreview-review |
Comment on attachment 8819524 [details]
Bug 1324193 - Bump rust-url to 1.2.4;
https://reviewboard.mozilla.org/r/99264/#review99616
Just a few minor questions that we can address upstream if needed.
Go ahead and land this.
::: third_party/rust/url/src/lib.rs:499
(Diff revision 1)
> /// URLs that do *not* are either path-only like `unix:/run/foo.socket`
> /// or cannot-be-a-base like `data:text/plain,Stuff`.
> + ///
> + /// # Examples
> + ///
> + /// ```
Not sure if this is required, but in the previous files you used \`\`\`rust
::: third_party/rust/url/tests/urltestdata.json:2164
(Diff revision 1)
> "hash": ""
> },
> {
> "input": "http://www.google.com/foo?bar=baz# »",
> "base": "about:blank",
> - "href": "http://www.google.com/foo?bar=baz# »",
> + "href": "http://www.google.com/foo?bar=baz# %C2%BB",
I was under the impression that spaces should also be percent encoded. No?
Attachment #8819524 -
Flags: review?(valentin.gosu) → review+
Assignee | ||
Comment 9•8 years ago
|
||
> Not sure if this is required, but in the previous files you used \`\`\`rust
Not required
> I was under the impression that spaces should also be percent encoded. No?
Not in hashes.
(That file is synced with the wpt testset, and that change comes from wpt)
Comment 10•8 years ago
|
||
Pushed by manishearth@gmail.com:
https://hg.mozilla.org/integration/autoland/rev/389bc37ce175
Bump rust-url to 1.2.4; r=valentin
Comment 11•8 years ago
|
||
bugherder |
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
status-firefox53:
--- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla53
You need to log in
before you can comment on or make changes to this bug.
Description
•