1377557 - Rewrite XMLHttpRequest/send-content-type-string.htm to comply with data: URI inheritance model

Status: RESOLVED FIXED

(Core :: DOM: Security, enhancement, P2)

Description

[Christoph Kerschbaumer [:ckerschb]]

Since you offered to rewrite the test I am assigning this one to you. For testing you can switch the pref |pref("security.data_uri.unique_opaque_origin", true);|

Thanks

Comment 1

[Christoph Kerschbaumer [:ckerschb]]

I just realized there would be 2 more tests that need similar rewriting:

> dom/nodes/Document-contentType/contentType/contenttype_datauri_01.html
> dom/nodes/Document-contentType/contentType/contenttype_datauri_02.html

Would you be willing to rewrite those two as well? Might fit into this bug :-)

Comment 2

[Anne (:annevk)]

I created https://github.com/w3c/web-platform-tests/pull/6470. Still needs review from someone. I ended up not rewriting the latter two tests as they were specifically testing the data URL part. Equivalent scenarios without data URLs were already covered. So I removed them.

Comment 3

[Christoph Kerschbaumer [:ckerschb]]

(In reply to Anne (:annevk) from comment #2)
> I created https://github.com/w3c/web-platform-tests/pull/6470. Still needs
> review from someone.

Thanks Anne. Can't we just land in our codebase and then upstream the changes?

> I ended up not rewriting the latter two tests as they
> were specifically testing the data URL part. Equivalent scenarios without
> data URLs were already covered. So I removed them.

That sounds good to me. But same here, can we remove them from our codebase then?

Comment 4

[Anne (:annevk)]

If that's better for you feel free to apply https://patch-diff.githubusercontent.com/raw/w3c/web-platform-tests/pull/6470.patch to the Mozilla tree (it's already been reviewed by jdm) and maybe squash the commits. I don't have a local checkout at the moment. If that's the way we go I'll close the PR and say we're addressing it here.

Comment 5

[Christoph Kerschbaumer [:ckerschb]]

Attachment: bug_1377557.patch

As per comment 4, took the patch from Anne and applied it to our tree. Carrying over r+ from jdm.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/401a95460411
Web platform tests update to comply with data: URI inheritance model: Update send-content-type-string.htm, contenttype_datauri_02.html and remove contenttype_datauri_01.html. r=jdm

Comment 7

[Christoph Kerschbaumer [:ckerschb]]

Attachment: bug_1377557_followup.patch

These [1] are the only 3 occurences of the testfile we removed. Do we have to do anything else, or just remove the file from manifest.json?

[1] https://dxr.mozilla.org/mozilla-central/search?q=contenttype_datauri_01.html&redirect=false

Comment 8

[Anne (:annevk)]

Comment on attachment 8883890 [details] [diff] [review]
bug_1377557_followup.patch

Review of attachment 8883890 [details] [diff] [review]:
-----------------------------------------------------------------

Looks alright to me, but not 100% I'm qualified to review this.

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/4d6404af7e75
follow up: Update manifest.json. r=me

Comment 10

[Christoph Kerschbaumer [:ckerschb]]

Comment on attachment 8883890 [details] [diff] [review]
bug_1377557_followup.patch

As discussed with jgraham on IRC, we should not manutally touch the manifest.json file, hence I ran |./mach wpt-manifest-update| and pushed the follow up to inbound. Here is the full IRC log:

ckerschb> Christoph Kerschbaumer padenot: if removing a test from wpt tests, is it enough to remove it from the manifest.json?
12:07 PM padenot: https://bugzilla.mozilla.org/show_bug.cgi?id=1377557#c8
12:08 PM 
<padenot> jgraham, ^ ?
12:10 PM ⇐ Conjuror and marcosc quit  ↔ cosmo0920 nipped out  
12:14 PM 
<jgraham> ckerschb: You shouldn't manually edit the manifest.json; you should just delete the file and run |mach wpt-manifest-update|
12:14 PM 
<ckerschb> Christoph Kerschbaumer jgraham: I see
12:15 PM jgraham: what happened is, that I already deleted the file and pushed to inbound
12:15 PM ⇐ breezykermo quit ([e-mail]) Ping timeout: 121 seconds
12:15 PM 
<jgraham> Assuming the test is wrong. If the test is right but we happen to fail it you should update the expectation metadata
12:15 PM 
<ckerschb> Christoph Kerschbaumer what do i do now?
12:15 PM 
<jgraham> ckerschb: Update the manifest using |mach wpt-manifest-update| and push a fixup commit to inbound

Comment 11

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/401a95460411
https://hg.mozilla.org/mozilla-central/rev/4d6404af7e75