[Biopython-dev] EUtils

[Andrew Dalke]

I was at the PyCon conference a couple weeks ago and talked with
Mark Johnson from NCBI about several things, including the EUtils
client package in Biopython.

That reminded me that I needed to clean up the package and update
it to support the latest NCBI interface (they added a couple new
features in the last two years).  I've got some free time this
week so I'm working on it.

The changes are:
   - don't require the response the DTD.  I now believe that DTDs
      for this task are the wrong approach.  I'm using elementtree.
   - get rid of the silly Problem class hierarchy I was using
   - simplify the regression tests so it's easier to read
   - add more tests (found some bugs in a few corners already)
   - figure out a better way to do live tests against the server
   - include support for a request throttle
   - pull the documentation out of the docstrings and into an
       actual document
   - resubmit my bug reports in the hopes that they'll fix
       the server
   - rearrange the source tree

For the last, what I want to do is move the documentation
and tests out of Bio/EUtils/ and into the proper places for
Biopython.  As stands EUtils is distributable independent of
Biopython.  I don't think that's worthwhile so I'm thinking
of no longer doing that.  If I do make it independently
distributable then I'll have some little script to assemble
the bits and pieces from the Biopython tree.

Is anyone here using the EUtils client for anything and
has code that I can use?  I want to examine common use
cases and see if there's a way to simplify the interface.
(Eg, have some top-level functions for doing retrieval
rather than making a client object.)  I also want to make
sure I don't change the API to break existing code.


					Andrew
					dalke at dalkescientific.com