[Biojava-l] Should we standardise on an XSLT processor?

Ewan Birney birney@ebi.ac.uk
Mon, 11 Dec 2000 15:45:21 +0000 (GMT)


On Mon, 11 Dec 2000, Simon Brocklehurst wrote:

> 
> An XSLT transform for converting between formats is certainly what we had in
> mind (but we haven't really looked in detail yet to see how much of a pain this
> will be).    Also, we've been holding off putting any XSLT stuff into biojava
> thus far because mechansims for adding Java extensions to XSL stylesheets aren't
> standardised.  It turns out that, for bioinformatics stuff, building Java
> extensions to XSL seems to be pretty much the rule rather than the exception.
> 

A quick $.02 on this. 


I am a big believer in XML (I want words with anyone coming up with a new
format not in XML...) and XSL for converting "trivial" format differences.


There is a rather murky and nasty grey area in bioinformatics which is
between "trivial format changes" and "data mining" which I think will tax
XSL implementations. There are times for example when you want to do
complex coordinate mapping between formats (think features on multiple
alignments) and call-outs to other id<->accession mapping tables etc.


In my view, lots of bioinformatics "format conversion" actually needs the
full richness of a language to handle all this ... challenging data ... 
and XSL will only buy you so much. Of course, this is an opinion, and I am
happy to be proved wrong.


Of course, the language can easily be java. (though I'd find Perl easier.
But with biocorba working, you can call out to my perl from your java
program...)



e.




-----------------------------------------------------------------
Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
<birney@ebi.ac.uk>. 
-----------------------------------------------------------------