xml-dev - Re: Parser Question.

[richard@cogsci.ed.ac.uk (Richard Tobin)]

{I had to cut short my previous message because the free ISP I use shuts
down at midnight.)

Brown, Bryan <bryanb@upshot.com> wrote:
>In the external subset PEs can occur anywhere, this seems to me to make the
>process of parsing an external production very complicated because I need to
>check at each step of the way if the next token is a PE reference, 

You can do it as part of whitespace processing.  Since a parameter
entity (outside a string) introduces a space at either end of its
expansion, a PE reference can only occur where whitespace can occur.

>The spec states that if a SystemLiteral has a fragment identifier the parser
>may signal an error, yet the conformance tests offer a document which is
>supposed not-wf because it has a SystemLiteral with a fragment identifier.
>So is it an error or is it the parser option what to do ?

Which test?  According to my notes, tests uri01 and o-p11pass1, which
have fragments in SystemLiterals, are marked "error" rather than "not-wf".

Anyway, the core WG is considering an erratum to clarify this.

>Is checking to make sure element content is deterministic an option or
>required for a conforming parser ?

It's not required, since non-determinism is merely "an error".  The
well-known implementation in terms of a finite state machine makes it
easy to check, however.

-- Richard
-- 
Spam filter: to mail me from a .com/.net site, put my surname in the headers.

"The Internet is really just a series of bottlenecks joined by high
speed networks." - Sam Wilson

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************