Off-topic. (Re: Generating a master set of schemas from the published Strict and Transitional schemas)

[Rick Jelliffe]

MURATA Makoto (FAMILY Given) wrote:
>>> I have not considered RELAX NG schemas yet.
>>>       
>> What is the current situation. Are they manually synchronized with XSDs
>> or are they generated automatically?
>>     
>
> Most of the original RNG schemas in 29500 were automatically generated 
> by XSLT, ruby, make, etc.  Others were written by hand from scratch.
>
> The converter from XSD to RNG was first written by Rick Jelliffe as an
> XSLT stylesheet.  I extended it significantly, and I added ruby scripts 
> and makefile.  The converter works but it is complicated and might 
> not be easy to maintain.  But if you are interested, I'm happy to 
> send the converter to you.
>
>   
>> The real question now is to whether I should learn F# or Scala. ;-D
>>     
>
> I don't think you have to.  You might want to concentrate on XSLT, 
> which is primarily for extracting S and T schemas from merged schemas.
> But you might want to edit merges schemas by hand and propose 
> changes to my F# program for merging.
>
> F# and Scala demonstrate type inference in real programming.  I enjoy it
> a lot.
>   
I think the biggest design problem with the original scripts was that 
they were based on document-by-document transforms: 1 RNG in = 1 XSD 
out. What I should have done is inserted all the RNG schemas in one big 
fat XML document first: that would have made it possible to keep 
entirely within XSLT2 I think: this is how I approached the XSD to 
Schematron converter, and it gets rid of much housekeeping code.  I 
think it would be neater to do that than to have the mixed strategy of 
XSLTs + scripts.

Cheers
Rick Jelliffe