[tei-council] Roma II

[Sebastian Rahtz]

Peter Boot wrote:

> As I wrote earlier, I have some doubts about the wisdom of re-developing 
> functional software.

In general, I agree. but Roma as it stands is
a bit of a monster, with a mixture of PHP, XSLT, XQuery,
Javascript, external binaries, etc, and a dependency on eXist running.
It is not very amenable to development, and is not
very easy to implement (as witnessed by the fact
that its not working right at Virginia)

> Why not put the effort in extending the existing 
> Roma with useful functionality, such as
> - adding copy element functionality,
> - supporting the creation/manipulation of content models (instead of the 
> user having to write the models),
> - support/help in creating schematron constraints?

The "copy element" thing is not that hard; the latter two
are pretty scary undertakings! they probably would themselves
be better suited to a pure JS solution? maybe not.

> I'd also say that other software (e.g. the TEI schema validation service 
> that there has been some talk about) really seems more urgent to me.

Indeed, you may well be right, which is why I raised this again.

The validation service is not that far off; see
http://rahtz.oucs.ox.ac.uk:8080/TEIISO/ for a proof of concept.
if that was given to a competent Java programmer for 2 weeks,
they could make it production ready for the TEI, perhaps.

> Could you explain what advantages there are in the proposed javascript 
> architecture? It doesn't seem to provide the desktop Roma environment 
> that people apparently would like to have.

if Roma was simply an HTML page with a huge amount
of Javascript, you could run it locally.

the main advantage is that it would remove the dependency
on a relatively complex server. you may not realize it,
but Roma does things like
    system("trang")
which is pretty danged scary from a maintenance and
deployment perspective.
-- 
Sebastian Rahtz
Information Manager, Oxford University Computing Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431