Re: Font Selection

[Danilo Segan <dsegan gmx net>]

Noah Levitt wrote:

>On Fri, May 16, 2003 at 12:49:41 +0200, Danilo Segan wrote:
>  
>
>>It will hardly happen, because it will invalidate and make many current 
>>web pages incompatible with that kind of browser. 
>>    
>>
>
>Yeah, but those pages are broken. They should be specifying
>their encoding. I've been using UTF-8 as the default
>encoding in my web browser for a while, and I mostly don't
>have problems.
>  
>
RFC 2616 quite clearly states in section "3.4.1 Missing character sets":

 > Some HTTP/1.0 software has interpreted Content-Type header without 
charset parameter incorrectly to mean "recepient should guess." Senders 
wishing to defeat this behaviour MAY include a charset parameter even 
when the charset is ISO-8859-1 and SHOULD do so when it is known that it 
will not confuse the recepient.

So, pages which don't send the charset parameter with value "ISO-8859-1" 
(if they use it) are "conditionally compliant" (because of the SHOULD, 
they're not "unconditionally compliant"). So, at the very least I 
wouldn't call them "broken".

By the way, I'm also using UTF-8 as default encoding for at least 2 
years, and I *did* experience problems when visiting such pages.


So, there're some problems with the tradition, but nothing which can't 
easily be fixed in the future (all my Apache installations have 
"AddDefaultCharset UTF-8" which "fixes" this, and the "official" Apache 
distribution comes with a sample containing "AddDefaultCharset 
ISO-8859-1" or similar, thereby helping those pages become 
"unconditionally compliant").

Cheers,
Danilo