W3C home > Mailing lists > Public > www-style@w3.org > April 2011

Re: Unicode normalization in CSS

From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Date: Mon, 11 Apr 2011 17:03:03 +0200
To: Daniel Glazman <daniel.glazman@disruptive-innovations.com>, fantasai <fantasai.lists@inkedblade.net>
Cc: Chris Lilley <chris@w3.org>, Henri Sivonen <hsivonen@iki.fi>, public-i18n-core@w3.org, www-style@w3.org
Message-ID: <20110411170303282711.3fa974ae@xn--mlform-iua.no>
Daniel Glazman, Mon, 11 Apr 2011 16:23:14 +0200:
> Le 11/04/11 16:11, Leif Halvard Silli a écrit :

>> My conclusion w.r.t. URLs, is that there is much need for advice. And
>> the advice should be split in two: [2]
>>   EITHER an author can gather info about which of "his/her" characters
>> that are affected by non-canonical issues. For example, in Norwegian,
>> with our ÆØÅæøå letters, only the 'åÅ' is affected. Likewise, in
>> Russian, only a pair of letters are affected.
>>   OR, to avoid the trouble, the author can get hold of tools that take
>> care of the normalization. In my case, I was finally able to find a FTP
>> application (Yummy FTP) which took care of the Unicode normalization
>> for me (I even managed to get the Yummy author to fix a bug). (But I do
>> wonder if Yummy FTP is the only Mac FTP programme that takes care of
>> this ...)
>> Ultimately, Apple needs to do (more) to fix the issue.
> Forgive me if I am a bit too harsh here, but I am not sure your
> answer helps a lot, Leif... We are not Apple, we're only the CSS WG
> and we try to understand what we should do here if we should do
> something. Reading the whole thread about unicode normalization in
> www-style, I am under the impression that the only conclusion is
> "this is a real and painful problem, and the experts of this problem
> clearly tell us CSS alone cannot solve it". Correct ?

Thanks for your remarks. 

First, your question: It is a problem. But mostly because it baffles 
people. A non-savvy author probably doesn't understand what goes on. It 
should not be left to CSS alone to solve all issues. But CSS should 
solve the issues that it can solve. I don't believe that one should 
look to UNICODE in the hope for an automatic salvation.

So, what should CSS do?

Looking at fantsai's letter [1], is her comments about CSS2.1 or about 
CSS namespaces or both?

One way forward for her could be to document, in a CSS specification, 
what letters that are affected. For example to authors, e.g. to 
Norwegian authors, it is not very logical that "æø" are not affected 
while "å" is affected.

Fantasai said she was in lack of conclusion. Since she is a spec 
editor, I offer the conclusion that we could document the issues. And 
by, "issues", I mean summarize/list the affected letters. And also to 
describe *when* it is likely to be an issue, such as when linking to 
files - which for instance affect a:visited{}. The testing I performed 
in connection with my letters to www-internatinonal [2] showed that in 
some of the tests in some browsers, the link might work while the 
a:visited{} style did not work. (When using percent encoding then the 
a:visited{} style was more likely to work than when not using percent 

That is: if the option 1. "Nothing is normalized in CSS" is selected, 
then the affected letters and issues should be documented. 

Even if CSS specifies that normalization is going to take place, there 
is still a need for documentation of the very letters that are 
affected. Since these issues are not related to only CSS namespaces, it 
does not need to only be in the CSS namespaces spec (except for those 
issue that potentially only affect CSS namespaces, such as evaluation 
of when namespaces are identical etc.)

[1] http://www.w3.org/mid/4D9E52A8.8060806@inkedblade.net


leif halvard silli
Received on Monday, 11 April 2011 15:03:44 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:38:45 UTC