[XOM-interest] Bug: XOM's rules for system ids are too stringent

Elliotte Harold erharold at gmail.com
Sat Feb 23 09:34:20 EST 2008


John Cowan wrote:
> (Private response)
> 
>> At first glance, this appears to be another incompatible change 
> 
> Quite right, but you're cherry-picking.  Can you point to *any*
> substantive errata that *don't* constitute an incompatible change?

Errata should not be used for making substantive changes to the 
language. That's what a new version is for. Errata should be limited to 
clarifying language and fixing outright mistakes. Frankly as long as a 
"mistake" is unambiguous, non-contradictory to other parts of the spec, 
and implementable, I don't think an erratum should be used to fix that 
either.

>> Thus the system identifier is a URI, not merely a string that is 
>> converted to a URI. It is an error for a URI to contain an unescaped 
>> linefeed, and XOM should throw an exception in that case.
> 
> You should document this as a major deviation from XML 1.0.  Or if you
> are implementing XML 1.0 2e, you should say so.
> 

XML 1.0 is deviating from itself. I obviously overestimated the ability 
or will of the W3C to maintain a stable spec with a consistent version 
number. :-(

-- 
Elliotte Rusty Harold  elharo at metalab.unc.edu
Java I/O 2nd Edition Just Published!
http://www.cafeaulait.org/books/javaio2/
http://www.amazon.com/exec/obidos/ISBN=0596527500/ref=nosim/cafeaulaitA/


More information about the XOM-interest mailing list