[Bioperl-l] Bioperl NT update

Peter Schattner schattner@alum.mit.edu
Thu, 14 Dec 2000 14:41:17 -0800

Hilmar Lapp wrote:
> Shailesh L Mistry wrote:
> >
> > The other minor problems relate to programs that I do not have on my machine
> > specifically blast, clustalw and TCoffee.
> > Do I assume that only people who use these programs should worry about the
> > testing of them or will they definitely be part of the distribution?
> >
> I guess they will, however many people will not use every part. Still, if
> you want to take the effort and get all those programs installed (as far as
> they are available for NT), that'd be great. I'm wondering whether the
> respective modules can live with a precomputed output, too. Peter?
>         Hilmar
The test scripts for the clustalw.pm, standaloneblast.pm and tcoffee.pm
modules check for the presence of the respective underlying programs
they need on the host machine when in a Unix (or Linux) environment.  If
they are not found, the respective tests are skipped.  However, under NT
this approach is untested and probably won’t work.  I suspect the
underlying programs may not even be available under NT (or Mac). 

I think the best bet would be to completely skip all the tests relating
to these modules unless the host OS is a Unix variant.  Alternatively,
the tests could presumably live with precomputed output to be used if
the Bioperl installation test-harness detects a non-unix environment . 
However I do not know how to incorporate either of these approaches in
the Bioperl installation.  Can anyone help?