Re: XSLT 2.0: Schema-aware processor: What are the co

At 09:33 AM 18/07/2007 +0100, you wrote:
>
>>From: Abel Braaksma <abel.online@xxxxxxxxx>
>>I don't think that Gestalt, "the other XSLT 2.0 processor" supports
schema awareness.
>
Colin Adams wrote:
>It does not.
>Although its architecture has been carefully designed so that it can act in 
>the future as either a basic or schema-aware processor, it does not look 
>likely that I will ever find the time to do it.
>
>But its an open source, open development project, so anyone can join in.

Wow, that's great to meet another XSLT processor developer.
May I ask you, Colin, what inspired you to take on such a task given
Saxon's high profile?

Perhaps I should ask myself the same question.  About 5 years ago I
developed a ~90% complete XSLT processor in C++ on top of my C++
persistence engine, ObjectAccess.
Since ObjectAccess supported automatic garbage collection, I was able to
run transforms over gigabyte size XML source files faster than any other
XSLT processor at that time .. all on a 386 box with 256K RAM.  I got a
bunch of huge XML data files from the Human Genome Mapping Project and felt
very satisfied when my system was able to process these.
Most XSLT processors either croaked or the (Windows) virtual memory  system
ground to a halt when the source got over several megabytes.  Saxon
survived the best out of all the others .. maybe its use of the tiny tree
helped it to conserve main memory ???

Anyway I had to make a living rather than persuing the XSLT hobby at the
time.  I ended up selling ObjectAccess to another company who subsequently
dusted it after decided relational DB rather than OODB was the new mood.
So that, together with seeing XPath 2 coming, I felt that too daunting to
take on.  For XSLT work I've used Saxon ever since (except in shops where
there's been a Microsoft mandate).

I never published the stuff on the Internet mostly due to a lack of time
and the perfectionist "disorder" that my wife says that I have.  I'll try
and change that this year :-)

So back to SA, would it be any easier to implement the XSLT beast using
RELAX-NG as the schema language rather than that W3C monster?

Justin Johansson
Freelance XML / XSLT / XQuery Developer
Australia

procode(at)tpg(dot)com(dot)au