Processing of &prod_id= in attributes from Adam Barth on 2010-06-29 (public-html@w3.org from June 2010)

From: Adam Barth <w3c@adambarth.com>
Date: Tue, 29 Jun 2010 11:29:28 -0700
To: Henri Sivonen <hsivonen@iki.fi>
Cc: HTML WG <public-html@w3.org>
Message-ID: <AANLkTil3CmeYCSfEmlMRlUPytBZuJpia9lTlPM-1OdCO@mail.gmail.com>

Hi Henri,

WebKit received a bug report [1] about URLs in attributes akin to the following:

http://example.com/foo?bar=baz&prod_id=qux

Recall that &prod; is an HTML entity.  According to my reading of
<http://www.whatwg.org/specs/web-apps/current-work/multipage/tokenization.html#tokenizing-character-references>,
we ought to consume the "&prod" in an attribute as a named entity.
However, Minefield does not appear to process the input stream this
way:

https://bug-41345-attachments.webkit.org/attachment.cgi?id=60037

Notice, however, that when we use &pound; in a similar way, Minefield
does consume "&pound" as a named entity:

http://example.com/foo?bar=baz&pound_id=qux

Is Minefield's behavior here intentional?  Should we update the spec
to explain why these two entities are treated differently?  Is my
interpretation of the spec incorrect?  (Note that the legacy WebKit
parser acts the same way as Minefield on these test cases.)

Thanks,
Adam

[1] https://bugs.webkit.org/show_bug.cgi?id=41345

Received on Tuesday, 29 June 2010 18:30:19 UTC