RE: [xml-dev] re: IDs and databases (Was: determining ID-ness in XML)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

True story.  The IDE/AS SGML hypertext system originally 
used PIs for link targets.  Because as an SGMLer at that 
time, I knew we would get beat up for using PIs, I convinced 
the developers to switch to attributes instead, and we even 
adopted the weakest nameloc hytime conventions we could. 
This made it over into IADS, the USAMICOM version.  We 
were pleased.  Then the SGML Way people began to beat 
the daylights out of us for requiring any kind of out 
of band information in the content because any DTD 
for IADS or IDE/AS had to support it.  We also 
required frame elements and made those optional, 
again, to stay out of the DTD.  Then we got beat 
up for not requiring a DTD (it was a well-formed 
system in the pre-XML days).  Sigh.... no win...

So the politics of markup can be ugly and religious. 
The PI will work.  The xml:id will work.  Rename it 
as xml:target or something like that, and it works 
better (no confusion).   But if we put it in the 
xml namespace, I think we owe it to everyone to 
be very clear about the requirement, the semantics, 
the expectation of support, and so forth, because 
the DTD writers can adapt and will but the 
processor developer needs very clear instructions.   

It is the requirement and the precise 
doctrinal decisions that make me pause.  We can 
add PIs as conventions until the troops come home. 
No problem. But I don't think that is what the 
proposers of this are after and until I see the 
whole proposal, I can't be comfortable with a 
fundamental expansion of the xml namespace.

As for putting PIs to sleep, don't kill a useful 
barnyard cat.  The rats may come back.  Preserve 
options.

len (paranoidly yours) 

-----Original Message-----
From: Champion, Mike [mailto:Mike.Champion@SoftwareAG-USA.com]

Likewise, untold generations of applications developers
would put their metadata  (e.g., information used to generate some UI
component on the fly) in a document as PI pseudo-attributes. I can kinda
sorta see how this could be done with namespaces, but it's exactly what PIs
were invented to do.  (Somebody please convince me otherwise; I would like
to see PIs put to sleep as much as anyone!).