As the above test case shows, the script element can be created with the html element as its parent. This apparently also exists in trunk, though I haven't tested this myself. I've occasionally seen link elements as a direct descendant of the html element, but I've never been able to consistently reproduce this (even with a single page) — possibly related?
Yeah, this does appear still in trunk, now I've actually tested it myself.
Occurs with Safari 2.0.4 (419.3) with original WebKit on Mac OS X 10.4.11 (8S165) as well.
Also tested with a local debug build of WebKit r30353 with Safari 3.0.4 (523.12.2) on 10.4.11.
What does HTML5 say?
Do you want me to paste the entire HTML5 parsing section into the bug? :-)
(In reply to comment #4) > What does HTML5 say? That <script> should behave like <base>, <link>, <meta> and <style> in the "before head" insertion mode and create an implicit <head>.
(In reply to comment #5) > Do you want me to paste the entire HTML5 parsing section into the bug? :-) > Smartass. :)
Created attachment 68063 [details] Patch
Comment on attachment 68063 [details] Patch Clearing flags on attachment: 68063 Committed r67857: <http://trac.webkit.org/changeset/67857>
All reviewed patches have been landed. Closing bug.
Comment on attachment 68063 [details] Patch Rejecting patch 68063 from commit-queue. Unexpected failure when processing patch! Please file a bug against webkit-patch. Failed to run "['./WebKitTools/Scripts/webkit-patch', '--status-host=queues.webkit.org', 'land-attachment', '--force-clean', '--non-interactive', '--ignore-builders', '--quiet', '--parent-command=commit-queue', 68063]" exit_code: 1 Last 500 characters of output: 68063&action=edit Fetching: https://bugs.webkit.org/show_bug.cgi?id=17023&ctype=xml Processing 1 patch from 1 bug. Cleaning working directory Updating working directory Processing patch 68063 from bug 17023. NOBODY (OOPS!) found in /Projects/CommitQueue/LayoutTests/ChangeLog does not appear to be a valid reviewer according to committers.py. ERROR: /Projects/CommitQueue/LayoutTests/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive).