Re: ISO-8859-3

Jukka K. Korpela wrote:

>> The validator validates, it can't debate.
> We can.

We also have no crytal ball telling us why an author might
wish (or be forced) to use "obscure" charsets.

We even have no clear definition of "obscure", but any
registered (unlike -11) existing (unlike -12) ISO 8859-n,
1 <= n <= 16, could qualify as "not exactly too obscure".

IIRC there was unfortunately no list of the top 100 most
popular charsets in Google's very interesting survey
<http://code.google.com/webstats/index.html>

Otherwise I'd say "grab top 100, remove junk, the rest
is a 'must have' for the validator".  And if some parts
of ISO 8859 don't make it add them anyway.

Especially if there's a corresponding registered charset
windows-mmmm with 234 identical code points as in 8859-m.

> Human beings (who can debate) have to decide which
> encodings are supported.

Anything users (plural) actually want and need is fine.
Some legacy charsets have nice properties wrt Unicode
like automagically "fully normalized".  It's also rather
simple to check if a device supports a legacy charset,
all of it, forever.  Try that with Unicode.

> Clearly it is easier to change the documentation.

But that makes no sense.  The code to support different
charsets is already there.  All it needs is a simple
table in this case, or has ISO 8859-3 strange features
not found in say -1 or -15 ?

Normally I tend to agree that all documented "bugs" are
"features", but it has to make sense.

A somewhat similar issue:  Why doesn't it support all
HTML DTDs published in RfCs like 2070 ?  It's no moving
target, there won't be any new HTML 2.x versions.

WDG can do, so why can't the W3C ?  Question posted in
<http://permalink.gmane.org/gmane.org.w3c.validator/7483>

>> It's not evangelize.w3.org, for that it has the tip of
>> the day.

> No, the "tip of the day" is there to confuse people.

Oops, I kind of liked it, maybe not all of the tips, but
many are at least interesting.  Okay, in essence it's a
gimmick.  OTOH folks like such gimmicks, compare STD 23:

          <http://tools.ietf.org/html/std23>

Make sure that your last will is up to date before you
try to move that full Internet standard to "historic" ;-)

                      Bye, Frank

Received on Saturday, 11 February 2006 16:22:13 UTC