Re: [topicmaps-comment] TMs & XTM [Was: skills to create topic maps]

[Robert Barta <rho@bigpond.net.au>]

On Thu, Jan 03, 2002 at 12:04:04PM +0100, Lars Marius Garshol wrote:
> The reason I keep calling the XSLT-based solutions half-baked is that
> to do things like reification, merging (which we really *do* need),
> and scope filtering in XSLT is so hard that nobody's going to do it.
> Which again means that we end up not making full use of topic maps,
> which is to nobody's advantage. XSLT is good for XML, it's not good
> for topic maps. A similar standard for topic maps would be enormously
> much more useful.
> 
> We could solve this by making a standard for this, based on TMQL, but
> that would mean waiting for quite a while. So what should we do?

We should wait. Seriously, I think that TMQL in one or the other
way _will_ have a facility to create queries resulting in

  (a) partial TMs which can be iterated/operated on via an API
  (b) directly XML code
  (c) ...

So, for example (plugging AsTMa?):

<ul>
  IN tm:opera
  WHERE SOME $t SATISFIES [ ? (opera)+ ] # select all topics being operas, directly or indirectly
  RETURN                                 # look at $t in XTM, of course fully merged
     FOR $b IN $t/baseName[scope/topicRef eq "#italian"]
        <li> {$b} </li>
</ul>

I doubt whether the nature of this XML structure can be defined
independently from the application.

Starting from here everyone is free to use XSLT but....

> We need to stop using XSLT and start using something better. 

....Lars is completely right: Using XSLT against XTM does not 'feel right'
(trademark). XSLT is about syntactic transformation. TMs are not
syntactical structures, IMHO.

Lars is also right, that we developers have to offer something like
the above ASAP.

On Thu, Jan 03, 2002 at 12:04:04PM +0100, Lars Marius Garshol wrote:
> I think you have this problem because you're using the wrong tools,
> not because of any problem with topic maps. To output an HTML <ul>
> list of all operas in a topic map, using their Italian names, sorted
> by those names, is about 10 lines using our XML language. 
....
> And the
> solution is not for everybody to buy Ontopia's software (even though
> that would be very nice for us in the short term).

In AsTMa? it will be 7 lines. Better wait.  8-)))

\rho