Re: follow-up on case sensitivity for user-defined identifiers in CSS

Hi John and CSS WG,

The I18N WG discussed this in our most recent teleconference also.

In some cases you examine, the WG is in agreement about case matching. There is little point to implementing a more complex match for e.g. case limited domains such as html tags.

However we don't agree that this is best policy. It was our opinion that this is a case where browsers will probably coalesce around whatever interoperable solution is adopted. Our concern is that this be understandable and appropriate for a global audience.

As it happens, I have been tasked with exactly the things you ask of us: providing a case for Unicode case fold matching. Alas I'm also traveling and won't finish this for a day or two, but I wanted to respond briefly now to say that.

Thanks,

Addison (for I18N)

Addison Phillips
Globalization Architect (Lab126)
Chair (W3C I18N WG)

Sent from my Kindle Fire HD

John Daggett <jdaggett@mozilla.com> wrote:


After reviewing the documentation of Unicode case matching, I think
those proposing "full" Unicode case matching for CSS user identifiers
are underestimating both the complexity of the implementation along
with it's utility.  The discussion at TPAC by the Internationalization
WG concluded that CSS should use full case matching for user-defined
identifiers but I don't see much discussion of the actual case
matching algorithm to be used.  I think the complexity involved
actually requires careful consideration and I've posted on
www-international stating the reasons. [1], [2]

I think the comment by Anne van Kesteren is most apt, for the problem
at hand full Unicode case mapping is overkill. [3]  Steve Zilles
summed this viewpoint up quite nicely during the telcon discussion
last week [4]:

> SteveZ: No strong opinions, but anne's solution more likely to be
>         interoperable.  Trying to track bugs for edge cases is bigger
>         pain than the value of being insensitive, and interop would suffer

If ASCII case sensitivity is seen as an inappropriate to use for
user-defined identifiers, I think we should stick with case sensitive
matching instead.  We should only opt for full Unicode case matching
if someone can desribe in detail the exact case matching algorithm and
justify why the complexity involved is worth the trouble.

John Daggett

[1] http://lists.w3.org/Archives/Public/www-international/2012OctDec/0141.html

[2] http://lists.w3.org/Archives/Public/www-international/2012OctDec/0142.html

[3] http://lists.w3.org/Archives/Public/www-international/2012OctDec/0109.html

[4] http://lists.w3.org/Archives/Public/www-style/2012Dec/0118.html