[topicmapmail] stupid question?

[Lars Marius Garshol]

* Bernard Vatant
| 
| Note that what is missing is *write* before XSLT ... which makes all
| the difference. And to this question, are there any free TM
| out-of-the-shelf stylesheets around? I guess the answer is no. 

The question is whether you need them, given that you can use the
Omnigator if you just want to browse your topic map. If you want a
customized interface, and you usually do, then you'll have to do
ontology-specific 'development' anyway.

| As Nikita explains quite well in the quoted chapter, you have to
| adapt your stylesheet to your TM specific structure, and singularly
| the templates (association types, role types, occurrence types) used
| in your TM.

You do, and if you are going to use XSLT/XQuery you also have to
restrict what XTM features you use. Basically you have to operate on
normalized XTM, because otherwise you will get into a horrible mess.
With topic map engines you don't have this problem, since they will in
a sense do the normalization for you.
 
| The layout is obtained through XSLT transformation of automatic XML
| output of Mondeca data base. The layout is generic, whatever the
| types of topic and associations, which is indeed a lazy and
| suboptimal solution.

Well, it's omnigation, which is useful as a first step. For this
particular project an ontology-specific interface would probably be
better, I agree.

| If you go to
| http://mondeca-publishing.com/s/anonymous/title10013.html, you can
| say it's quite messy and I will agree. The generic stylesheet gives
| a poor rendition of information here and there. 

I think there are two primary reasons for that: overuse of n-ary
associations and the fact that you use XSLT. Generic association
presentation is difficult, and the reason the Omnigator can do it so
well is because the layer underneath is designed for topic maps,
making it easy to express the presentation you want.

| That's another reason why we need libraries of templates for
| widespread simple (I won't say standard) association types, like
| e.g. those expressed in Thesaurus, or in Dublin Core metadata ... 

I agree. Having publicly available cookbooks of topic map patterns
would make it much easier for people to get started on their own topic
map designs. I've been thinking the same thing many times, but so far
not had time to do anything about it.

| If we had that stable templates, smart XSLT developers could provide
| out-of-the-shelf stylesheets allowing non-technical, non-fortunate
| (or simply lazy) people to publish their TM at low cost, and without
| going through the hassle of specific XSLT development. We're not
| there, but that's a way to go.

I very much doubt that this will actually work, and with a topic
map-specific XSLT-like language you don't need it, either. Traversing
associations and displaying topics then becomes so easy that you have
no need for a library that can do the job for you. *That* is what we
really need, so that people can forget about using XSLT for something
it is eminently unsuitable for.

This website, for example:
  <URL: http://www.ontopia.net/i18n/index.jsp >

was published using a topic map-specific publishing language, and
because of that it was very easy to develop and give a reasonable
presentation to. (The defects in the presentation are due more to my
lack of skill as a user interface designer than to anything in topic
maps themselves or the underlying language.)

The conclusion to this is that what we need is a TMQL in at least two
parts: one for simple querying (XPath++) and one for textual output
(XSLT). Only when we get open source implementations of this will
people like Nikita and Suellen have real tools to work with that will
make the full power of topic maps available to them.

I know I've said this before, but people *really* need to stop
thinking XSLT is a real solution for topic map publishing.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >