518104 - [HTML5] Implement HTML5 changes to <script defer>

Status: RESOLVED FIXED

(Core :: DOM: HTML Parser, defect)

Description

[Henri Sivonen (:hsivonen)]

Need to change <script defer> so that defer is ignored on inline scripts and document.write() for an external defer script blows away the document.

Comment 1

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Attachment: WIP

This fixes the code, but we need to also change tests that test of the old behavior. Henri has graciously offered to do that and drive the patch into the tree.

Comment 2

[Henri Sivonen (:hsivonen)]

Comment on attachment 404070 [details] [diff] [review]
WIP

Looks good to me, except it seems that the behavior is correct per HTML5 only if another script doesn't go and change the src attribute between the time the parser inserts the node and the time the parser parses the end tag. However, that problem is pre-existing and not introduced by this patch. (Unless I'm missing something about how GetSrc works.)

Comment 3

[Henri Sivonen (:hsivonen)]

Attachment: Trunk patch with test tweaks

Comment 4

[Henri Sivonen (:hsivonen)]

Attachment: Branch patch with test tweaks

Comment 5

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 404228 [details] [diff] [review]
Trunk patch with test tweaks

r=me on the test changes

Comment 6

[Henri Sivonen (:hsivonen)]

Attachment: Trunk patch with test tweaks, v2

Comment 7

[Henri Sivonen (:hsivonen)]

Attachment: Branch patch with test tweaks, v2

Comment 8

[Henri Sivonen (:hsivonen)]

Per discussion with sicking, getting this on the branch would be good, since the longer Gecko is HTML5-incompliant on this point, the more likely it is that Web content starts depending on Gecko's old behavior.

Comment 9

[Henri Sivonen (:hsivonen)]

Attachment: Trunk patch with test tweaks, v3

Comment 10

[Henri Sivonen (:hsivonen)]

Attachment: Branch patch with test tweaks, v3

Comment 11

[Henri Sivonen (:hsivonen)]

Try server id f1b91fc7f1b9.

Comment 12

[Blake Kaplan (:mrbkap) (inactive)]

Oh, we probably need to leave the functions in to be compatible with the older branch.

Comment 13

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Landed on trunk and 1.9.2

http://hg.mozilla.org/mozilla-central/rev/1db61f2dfded
http://hg.mozilla.org/releases/mozilla-1.9.2/rev/170e05103f8b

Comment 14

[Jesse Ruderman]

This bug is kinda the opposite of bug 461555, which was blocking1.9.1+ because we were aware of several web sites that it broke.  Are those sites broken again? :(

Comment 15

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

The hope is that they are not. All of the examples that I looked at that broke in that bug was due to inline scripts. Inline scripts are no longer deferred so it should be fine to document.write in them.

Comment 16

[Anthony Ryan]

The implementation of this change has reopened another bug https://bugzilla.mozilla.org/show_bug.cgi?id=474392

Comment 17

[What.CD]

I have put together a demonstration page of this functionality breaking backwards compatibility and a number of test cases which are generally accepted to work in all browsers.

http://what.cd/defer-bug.php

Comment 18

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

What.CD: I don't actually understand what your page is saying, but it's not related to this bug, so please file a separate bug on this.

Or better yet, bring it up on the WhatWG mailing list as it appears to me that you are disagreeing with what the spec says rather than you have found a bug in our implementation of the spec.