xml-dev - Re: Common XML (was Re: Document Feature Requirements)

[Rick JELLIFFE <ricko@geotempo.com>]

"Simon St.Laurent" wrote:
 
> Beyond that, we're not claiming that comments or processing instructions
> should be 'part of the document structure'.  What we're saying is that
> developers who want to make sure that this information is preserved across
> multiple levels of processing will probably have better luck if they create
> element structures that perform similar tasks.
 
This is what I don't get. Everyone knows that you don't put data in
comments. 

XML Spec at 2.5 is clear on it.  Anyone who knows HTML history or writes
JavaScript knows that SSI's use of comment delimiters and Netscapes use
of <!-- inside SCRIPT (and STYLE?) elements also abuses what comments
do. Everyone who has used a programming language knows that comments
don't carry program data but information related to program as source
code. If I compile a program and then disassemble it, I don't expect my
comments to be intact.

If I send my document through XSLT I am happy that the default behaviour
is to strip comments: I am generating a new text (new "source code" so
there is no reason to expect the comments to be appropriate, by
default). I don't want a comment "oops this is a terrible hack, the
foo:squidget  has to many blort:zzzz
subelements" to appear in generated HTML code: it is irrelevent.

Have I missed the rise of a group of madmen who, distrusting elements,
want to only use comments for passing data around?  Why is there a need
to make up a problem and then warn about it?   It is a little like
warning that the use of the number 0 has bad reliability problems
because when you divide by it many
applications will break: this is no reason to say  0 is "non-core" with
regard to reliability.  Just use it appropriately.

I think there is a difference between "reliable" and "core/peripheral".
The latter is defined well by the W3C infoset document: comments are
peripheral to the information set. As for what is reliable, it seems
that "round-trippable" is being used as the measure, but it is not the
same as reliable. CDATA marked sections do not need to be
round-trippable to be reliable.

It mars what otherwise could be a useful and laudable document.  A
comment is not something that "makes it difficult to ensure that varying
applications will receive the same view of a document."  


Rick Jelliffe

(Please note that Simon is also persuing, through the Web Standards
people, attempts to make developers support standards fully. So I hope
no-one thinks that he is ignoring the solution of making developers
support the XML Spec properly. To use an Australian expression, he is
walking both sides of the barbed-wire fence.)

***************************************************************************
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/
***************************************************************************