re: PSVI

["Simon St.Laurent" <simonstl@simonstl.com>]

At 05:45 PM 3/1/01 -0500, David Megginson wrote:
>My preference is to include all defaulted information in production
>instances and not to include a DOCTYPE declaration or schema link at
>all -- that's the only way to ensure that all clients see the same
>document, and that no one does anything stupid.

Generally speaking, I think David's right, though there are at least two 
options which permit the use of DTD defaulting without the security risks, 
inflexibility, or other potential hassles.

a) rely on PUBLIC identifiers mapped to a local cache.  Treat the system 
identifier as merely a convenience for the sender, but require PUBLIC 
identifiers which match to a list of DTDs maintained by the program.  See 
http://xmlhack.com/read.php?item=392 for information and Arbortext's 
implementation.

b) modify the DOCTYPE declaration of incoming documents to reflect a type 
your program expects, and point it at a local resource.  I've written a 
DOCTYPEChanger filter in Java which does this - 
http://simonstl.com/projects/doctypes/.  It's not that difficult.

Both of these allow you to use DTD-based defaulting without relying on 
external DTD resources and the kindness of senders.  Neither one is widely 
used in practice, from what I've heard from people, but they can be helpful 
at times.

(And instead of Infoset vs. PSVI, I'd love people to start thinking in 
terms of SI - standalone [no external resource] infoset, DTDI - 
DTD-influenced infoset, and then PSVI.  Brought to you by the Committee for 
Unusable Acronyms, CUA.)


Simon St.Laurent - Associate Editor, O'Reilly and Associates
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books