W3C home > Mailing lists > Public > public-i18n-core@w3.org > January to March 2009

Re: [CSS21][css3-namespace][css3-page][css3-selectors][css3-content] Unicode Normalization

From: Boris Zbarsky <bzbarsky@MIT.EDU>
Date: Mon, 02 Feb 2009 10:24:56 -0500
Message-ID: <49871048.8070505@mit.edu>
To: Andrew Cunningham <andrewc@vicnet.net.au>
CC: public-i18n-core@w3.org, www-style@w3.org

Andrew Cunningham wrote:
> i was saying that it is dependant. certain aspects, such as matching
> selectors require normalisation. But wholesale normalisation of everything
> will not be appropraite.

The issue is performance.  Right now, matching a tag name selector in 
Gecko is a single pointer comparison.  If I understand your point of 
view about form data correctly, then any data that can be communicated 
to the server must not be stored in normalized form, yes?  That 
basically means all CSS and DOM data.

That makes the comparison of tag names, say, much more complicated and 
much slower than a simple pointer comparison.  I realize that the simple 
case can be fast-pathed, but you're still introducing a good bit of 
overhead; at the very least you have different storage for data that can 
and cannot be affected by normalization and a runtime check for your 
storage types...

Or did I misunderstand the constraints on normalizing data in general?

-Boris
Received on Monday, 2 February 2009 15:25:42 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 2 February 2009 15:25:43 GMT