[SCL] some explanation

[Rich Morin]

At the risk of getting in WAY over my head, I'd like to make a small
comment.  Please don't beat me up; I'm just trying to help...

XML is rather like a glacier, slowly but inexorably pushing its way
into the landscape.  So, I agree that any standard for the exchange
of information should have an XML binding.

That said, XML is not a particularly good way to serialize all types
of data structures.  After loading in an XML tree, I generally have
to perform quite a bit of post-processing before I can navigate it.

So, I tend to use YAML (www.yaml.org) as my "native" serialization
format.  It is easy to read and a good fit for Perl, Python, Ruby,
etc.  I can load in a YAML file and immediately start using the data
structures (eg, hashes and lists) it encodes.

Similarly, I suspect that some of the alternate encoding formats for
SCL are much better matches than any XML-encoded format for the way
the information will be used in a given processing system.  It would
therefore be rather silly to require these systems to read XML, then
post-process an enormous set of data structures before using them.

I'd also like to point out that XML is ONLY a serialization format.
Trying to use XML as an alternative to an RDBMS, let alone graphical
data structures, seems misguided, at best.

So, assuming that SCL's semantic and structural information can be
mapped into XML (however ungracefully), do so.  But don't make XML
the definitive serialization format unless it matches your needs.

-r
-- 
email: rdm at cfcl.com; phone: +1 650-873-7841
http://www.cfcl.com        - Canta Forda Computer Laboratory
http://www.cfcl.com/Meta   - The FreeBSD Browser, Meta Project, etc.
http://www.ptf.com/dossier - Prime Time Freeware's DOSSIER series