W3C home > Mailing lists > Public > public-html-bugzilla@w3.org > June 2010

[Bug 9843] New: Specced behavior for document.write("<link rel=stylesheet href=...><script>...</script>...") matches none of the top 4 engines

From: <bugzilla@jessica.w3.org>
Date: Thu, 03 Jun 2010 13:07:53 +0000
To: public-html-bugzilla@w3.org
Message-ID: <bug-9843-2486@http.www.w3.org/Bugs/Public/>
http://www.w3.org/Bugs/Public/show_bug.cgi?id=9843

           Summary: Specced behavior for document.write("<link
                    rel=stylesheet href=...><script>...</script>...")
                    matches none of the top 4 engines
           Product: HTML WG
           Version: unspecified
          Platform: PC
               URL: http://hsivonen.iki.fi/test/moz/sheet-blocking-script.
                    html
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: HTML5 spec (editor: Ian Hickson)
        AssignedTo: ian@hixie.ch
        ReportedBy: hsivonen@iki.fi
         QAContact: public-html-bugzilla@w3.org
                CC: bzbarsky@mit.edu, mike@w3.org, public-html@w3.org,
                    jonas@sicking.cc, eric@webkit.org


Consider:
document.write("<link rel=stylesheet href=...><script>...</script>...")

According to the spec, the style sheet becomes a style sheet blocking scripts.
http://www.whatwg.org/specs/web-apps/current-work/multipage/semantics.html#a-style-sheet-blocking-scripts

The according to
http://www.whatwg.org/specs/web-apps/current-work/multipage/tokenization.html#parsing-main-incdata
, case An end tag whose tag name is "script", subcase "Otherwise", step 3, the
UA must "Spin the event loop until there is no style sheet blocking scripts and
the script's "ready to be parser-executed" flag is set."

As far as I can tell, this is not what happens in reality. What seems to happen
is:

In Gecko (old and new parser), the style sheet becomes a style sheet blocking
scripts, the parser blocks at the written </script> and document.write returns
early before the written content after </script> has been tokenized.

In IE8, the style sheet becomes a style sheet blocking scripts, but
document.written internal scripts are immune to style sheets blocking scripts,
so the document.write tokenizes to completion and the next inline script from
the network blocks on the style sheet. (Sorry, the demo doesn't prove the part
about the next network-originating script blocking without editing the demo and
experimenting.)

In Chrome beta channel, the style sheet doesn't block scripts but blocks
painting. The whole page is parsed to completion but isn't painted until the
style sheet has been loaded.

In Opera, the style sheet doesn't block anything. The page FOUCs.

I think *any* of these four behaviors is preferable over creating a nested
event loop from within document.write(), so I request the spec be changed to
one of the pre-existing behaviors on this point.

If Gecko's behavior is chosen for the spec, it would be necessary to specify
that the style sheet fetch completes asynchronously (even if the style sheet
were available immediately) in order to avoid making the document.write return
behavior sensitive to caching or to the URL scheme (data: URLs).

I don't have data to be able to argue the choice among the four behaviors based
on compatibility data. IE's behavior is the hardest to distinguish from what
the spec says now.

Note that it's not necessary to change the behavior for the case where the
<link> and the internal script come from the network stream, since that can be
handled per current spec (and Gecko does) without creating a nested event loop.

-- 
Configure bugmail: http://www.w3.org/Bugs/Public/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
Received on Thursday, 3 June 2010 13:07:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 3 June 2010 13:07:55 GMT