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

Re: Hoisting <base> into <head>

From: Henri Sivonen <hsivonen@iki.fi>
Date: Sun, 12 Sep 2010 04:20:04 -0700 (PDT)
To: Jonas Sicking <jonas@sicking.cc>
Cc: public-html <public-html@w3.org>
Message-ID: <142058576.548986.1284290404670.JavaMail.root@cm-mail03.mozilla.org>
Jonas Sicking wrote:
> Personally I'm not convinced that this breakage is large enough to
> warrant changing the spec. Not hoisting elements around certainly
> keeps the spec and implementation simpler so that's IMHO a good goal.

We've changed the spec to address one-off breakage on less prominent prominent sites than united.com, and the Hypelatex breakage affects multiple sites. (And on the topic of pragma no-cache in meta we even have a bug open without data of any specific real site breaking.)

I think in this case the spec change to hoist to <head> would be super-simple: One "process according to the rules the foo mode" case. The implementation would add a few lines of code but nothing dramatic.

However, I'm starting to prefer leaving the parser unchanged and taking into account the first <base href> (or <base target>) in the document. This would be a true spec simplication: Saying "first foo in the document" is simpler than saying "first foo that's a child of bar". Looking at the Gecko implementation of <base href>, it seems to me the implementation could become a couple of lines simpler or stay at about the current level of complexity.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Sunday, 12 September 2010 11:20:39 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:17:14 GMT