No to RELAX-NG (for now)

Thursday, January 03, 2002


XML-Dev,

Let me give you a practical viewpoint on XML standards and schemas in
particular.

The big problem with XML is that these floating standards never
get out of committee.  Now we see schemas finally making their
escape.  That is a good thing.  I dislike the idea of replacing
them with something else.  That move is like taking a 90-degree turn in
the road at 60 MPH.  XML will crash.

Because schemas have been so long coming, everyone I meet is using DTDs.
This kind of backwardness is caused by uncertain, zig-zag standards development.
Nobody wants to commit to the 'new thing' because at least the 'old
thing' is stable.  I have participated in XML design meetings where
this argument won the day.  The participants admit that yes, this or
that is better, but hey -- this is what the committee has standardized
and what industry supports, so we will use the ugly old stuff instead.
That is how industry works and it's important for XML folks to
understand that fact just as well as they understand trees, mappings, hash
functions, and relational tables.  Microsoft is so successful
precisely because they understand this fact so well.  That is why I
can run DOS programs under Windows 2000.

I have edited schemas with XML Spy.  No problem.  I understand them.
It was a bit like learning a new language, but still a very mild experience.
I have no problem editing schemas by hand, either, and XML Spy
supports that mode too.

I grant that RELAX NG may be better.  Frankly, I've never heard of it
until now.  I looked at the web sites.  Ho hum is my impression --
more XML tweaking when what the world needs is a stable XML standard.

It's not that I particularly love schemas (though much better than
DTDs).  It's just that I have seen this downward slide before.  Example:  while
Microsoft plodded along with its notoriously inept Microsoft
Foundation Classes, no one else offered a stable alternative, so MFC
became the standard for Windows development.  Notice I said, "stable"
alternative.  The same arguments won the day on those battlegrounds, too:
"this is what industry supports."


Kind regards,

__________________________________________________________________
Mark Evans              Control Systems Engineer
Zyvex Corporation       http://www.zyvex.com/
__________________________________________________________________

Thursday, January 03, 2002, 3:55:15 AM, you wrote:


>>[Jonathon Robie]
>> > RELAX-NG is a simple and straightforward language, but it is defined with
>> > formalisms quite similar to those of the XML Query Formal Semantics. XML
>> > Schema is not as simple and straightforward as RELAX-NG, and this may be
>> > partly due to the fact that its formal semantics were defined after the
>> > language itself

>>[Michael Champion]
>>RELAX-NG is getting good buzz these days not because it's based on the
>>formalism(s) of hedge automata and tree regular expressions, but because
>>it's elegant -- simple yet powerful. RELAX/TREX  are elegant because Makoto
>>Murata and James Clark very deeply understand both the underlying formalism
>>and XML itself. No amount of post-hoc formalism can create elegance when it
>>does not exist in the core of a design.

SM> Its also getting buzz because you can fire it up in Emacs or Windows Notepad
SM> grok it, change it and so on. You don't need any GUI tool. This is not to say
SM> that GUI tools are not desirable for RelaxNG. The critical issue is that they
SM> are "nice to haves" not "must haves". When we get into visual
SM> tool "must haves" as we are in some ares of XML standards work we are
SM> in dangerous waters because the visualizations are not part of the 
SM> standards - therefore
SM> vendors create different visualizations. If the visualization *becomes* the 
SM> notation, no
SM> amount of XML syntax for persistence will save you from the subltle, slow 
SM> squeeze
SM> of embrace and extend.

SM> Sean