Re: [rng-users] Lets standardize PI for associating Relax NG schema with XML document
- George Cristian Bina wrote:
> The main reason for having a PI is the *portability*.Portability? In user examples I've seen, most schema location URIs are
either at well-known network locations, e.g., web-xml, or file: URIs
that aren't the least bit portable.
> Also the external association cannot handle for instanceWhy not?
> cases when a user wants to specify that a document should be associated
> with a lite version of the main schema.
> Tools will always have the possibility to provide a "Validate with..."Exactly, so the PI isn't needed.
> action and allow the user to specify the schema externally, or to use
> the external association rules.
- MURATA Makoto (FAMILY Given) wrote:
> If the camp trying to standardize PIs is not the majoritySorry to answer so late, I've been on vacation. Just for the record, if
> of the RELAX NG community, I do not think that PIs will take off.
> Here is my understanding of the current status. Please let me know
> if I misinterprets somebody.
> For schema-associating PIs
> Jirka Kosek
> Robin Berjon
> George Cristian Bina
you are going to be counting heads in the RNG community, I think I would
be best counted as "neutral". My take on this is that a schema PI is
just as bad an idea as a stylesheet PI, which is to say that it's most
of the time a very bad idea (and in the absence of a processing model,
dreadfully underspecified in its interactions with other specs at that),
but *if* people are going to be doing it anyway (as seems to be the
case) then I would prefer that there is a standard made by people who
understand the issues and limitations of this approach rather than ad
hoc proprietary options mades by people who are probably smart and
probably understand some of the problems, but won't benefit from the
head-banging that some form of community standard would get (or rather,
Senior Research Scientist