Re: what's a parser's job? [was: RE: SD5 - Namespaces - New Version 2]

The XML-Lang spec calls the parsing thingie an "XML processor."  And there
are two kinds: nonvalidating and validating.  Are these unclear with
respect to classic CS terminology?

	Eve

At 10:09 PM 5/23/97 EDT, lee@sq.com wrote:
>Andrew Layman <andrewl@microsoft.com> wrote:
>[...]
>> 	[Andrew Layman] This is the key question here. It is not the XML
>> parser's job to detect namespaces. Applications that care will break out
>> the namespace part of the name, and look for a matching namespace
>> element. This could be a standard support routine shipped with parsers,
>> but is not--strictly speaking--a parser's job. Similarly, I do not see
>> validation as part of a parser's job, but as something that is (at least
>> logically) layered on.
>
>In the SGML world, the word "parser" does not mean "parser" as used
>elsewhere in the field of computing.  It has a special, SGML meaning.
>An SGML parser does indeed do validation.
>
>Whether an XML parser (in the usual sense of the word parser) should do
>validation is another matter entirely.
>
>I agree with you that namespace checking doesn't belong in a (CS) parser.
>It's part of the semantics layer, if you will.  But I wouldn't expect a
>CS parser to check that an element had been declared either -- that would
>be done in a separate module/class/layer.
>
>In this mailing list at least, the word "parser" is generally taken in
>the SGML sense, although I would agree that it isn't clear.
>
>Would it help if the XML spec used different terminology?
>E.g. the XML Input Reader and the XML Input Checker, or something??
>(an SGML/XML parser contains both of these)
>
>Lee
>
>
>

Received on Saturday, 24 May 1997 15:44:01 UTC