[Biojava-l] New Das-client release

Thomas Down td2@sanger.ac.uk
Fri, 7 Sep 2001 12:46:21 +0100


Hi...

I've just put out a new release of the BioJava das-client
browser (dated 20010907).  You can download it from:

   http://www.biojava.org/download/das-client/

There are a few instructions at:

   http://www.ensembl.org/das/client.html
   
There have been quite a few changes under the hood
since the last release:

   - Major work on the network layer (most if it for
     the good :-).

   - Much tighter error-handling in the client, including
     the option to recover from many kinds of fault.

   - Improvements in the handling of multiple servers
     in the sequence-chooser.

   - Uses the latest version of Eric Albert's BrowserLauncher.java,
     which should work on MacOS X.

   - Added a `stylesheet guesser' (currently rather simplistic)
     for servers which don't have stylesheets, or which supply
     a seriously invalid stylesheet.


This release has been tested with the following server
implementations:
 

    Dazzle (0.9x):  (Using either DAS/0.999 or DAS+XFF)
                    Human genome @sanger, plus many annotation
                    servers.  Should also work with the
                    Drosophila Dazzle (what's the status of this
                    now?)

    UCSC:           Human genome

    Wormbase.ORG:   C. elegans reference server.

It's nice to see a good level of interoperability between
these implemetations: congratualations are due all round!



Known issues:

   - The chromosome outliner currently only works with the
     Ensembl data.  However, you no longer need to explicitly
     turn it off before connecting to UCSC.

   - Wormbase currently has a stylesheet with several undocumented
     glyph types.  You'll get an error popup warning about this,
     but if you click `continue', the stylesheet guesser will come
     up with something reasonable.

   - The wormbase server has a known issue whereby the `types'
     command fails messily for large segments.  There's
     currently a workaround for this in the network code, but
     it's quite high-risk, and I'd like to get rid of it
     quickly.

   - Rendering performance isn't as good as I'd like it to be :-).


Enjoy!

   Thomas.