Re: [gobo-eiffel-develop] RNC file for geant?

[Colin P. A. <co...@co...>]

>>>>> "Eric" == Eric Bezault <er...@go...> writes:

    Eric> Colin Paul Adams wrote:
    >> Yes - it is to uniquely identify the namespace, so that the
    >> document element's system and library can be identified as the
    >> XACE ones, and not someone elses.

    Eric> And why do we need to do so? In an Xace file it looks
    Eric> obvious to me that system and library are the Xace ones.

Well, we don't absolutely need to do it. But it is a good idea to do
so, as it eliminates ambiguity.

I asked for it, because nXML mode identifies the RELAX-NG schema to
apply to a file by a series of rules. I am have to use the document
element to find the correct schema. But this could very easily mistake
some other XML file for an XACE file - after all, system and library
are not exactly rare words.

Now this isn't the end of the world it this were to happen - I would
just have to adjust manually. But I think the scheme will bvery likely
spread to other applications of RELAX-NG. 
Unlike the case of W3C schemas, XML files have no method to indicate
the schema that should be used to validate them. Since James Clark
came up with this particular scheme, you can probably risk your
savings betting on it's wider adoption.
Now, when we finally write a RELAX-NG validator, I think we will want
to use this kind of scheme. And we can deliver a pre-canned rules file
within Gobo. But if it is matching on the document element local name,
rather than a namespace URI, it won't be very reliable.

    Eric> So, are you saying that instead of having:

    Eric> <system> ...  </system>

    Eric> we would have this less readable (because more verbose)
    Eric> version:

    Eric> <system xmlns="http://www.gobosoft.com/schema/xace/1"> ...
    Eric> </system>

Yes.
-- 
Colin Paul Adams
Preston Lancashire