Re: Unrecognized encodings (was Re: XML 1.0 Conformance Test Results)

[Tim Bray <tbray@textuality.com>]

At 11:12 AM 11/06/01 -0600, Mike Brown wrote:
>My question is, must the XML parser developer honor these "shoulds" as if
>they were "musts" and produce a fatal error rather than accepting "UTF8"?

Hmm, in an earlier pre-REC draft of XML 1.0, there was a little 
homily about why processors should be forgiving on this issue.  
There are 2 possible errors: (a) you fail to parse something because 
you don't know the encoding, and (b) you guess wrong at the encoding, 
with unpredictable results but *probably* fail to parse.  Since (b)
- guessing wrong - is highly unlikely, (a) is probably worth 
doing some work to avoid.  Anyhow, the only leftover of all this
is the word "should".  In any case, I'd write software to accept
UTF8, but I'd complain at anyone who sent me data so labeled. -Tim