[Bioperl-l] Bioperl 1.5.2 RC2

[Chris Fields]

> I've just added test results for 1.5.2 RC2 to the wiki.
> 
> There are lots of fails for packages other than bioperl-live. I'm not
> sure excatly how the test fails/skipps are/should be handled since my
> setups are as follows.
> 
> Clean WinXP Pro:
> This is a clean install of WinXP Pro SP2 with no major software
> installed, other than ActivePerl 5.8.8.819 and a few tools for archive
> extracting, anti virus etc. Therefore, I'm unsure how tests in
> bioperl-network and bioperl-db should return. For example, I have made
> no effort to setup biosql-schema but I thought that maybe there would be
> a test that would detect this, and fail, then skip over other tests
> gracefully - like the bioperl-run tests when a piece of software is not
> installed???
> 
> Debian Linux:
> This is a Bio-Linux machine with quite a lot of bioinformatics software
> installed in the Path. So most of the tests in bioperl-run should
> probably have passed. The same goes for bioperl-network and bioperl-db
> as with my Windows setup.
> 
> If my thoughts are totally wrong - let me know!
> Nath

The bioperl-db tests rely on a local BioSQL database and on having a
properly set up configuration file (these are detailed in the bioperl-db
INSTALL doc).  Furthermore, there are serious problems with bioperl-db and
WinXP (see Bug 1938 in bugzilla).  There is a workaround, but it isn't
perfect by any means.  

http://bugzilla.open-bio.org/show_bug.cgi?id=1938

Many of the bioperl-run tests rely on env. variables being set properly, so
maybe that's why they failed.  These should all be detailed in the INSTALL
file (but maybe they aren't?).

I haven't tried installing bioperl-network or bioperl-run on WinXP or Mac OS
X yet but intended on doing this within the week.  The INSTALL file details
the requirements for the packages (Graph 0.80 is the only one for
bioperl-network, for instance, and there isn't a PPM for that version
available yet).  

It would be nice to skip the tests based on absence of the particular
modules or installed programs, and I think the final goal is to possibly
attempt to do this.  However, all of the bioperl-related distributions have
their own documentation which outline their installation, requirements, and
use.  At least we can point to that, which works for now.  We could always
start up a wiki page for the various bioperl distributions to monitor
problems or issues with each based on OS, proposed enhancements/ideas, etc.


Also, most (if not all, including core) have been primarily tested on some
*nix-related system, which means that they may not work on Win32 systems.
Though the Windows support is light-years ahead of what it used to be circa
rel 0.7, I don't think it is full-proof yet, as witnessed by the bioperl-db
bug.  Frankly, we need more WinXP users for those packages willing to test
them out and offer suggestions.  

Christopher Fields
Postdoctoral Researcher - Switzer Lab
Dept. of Biochemistry
University of Illinois Urbana-Champaign l