[Bioperl-l] tests failing miserably

[Jason Stajich]

I've added some changes to Bio::Root::IO which allow one to explicitly
request the IO object to not auto-close a filehandle.

The non-closing fh boolean flag was added to fix problems with the
Index::Blast object b/c we were using a shared filehandle but delegating
it to the a parser to get back a Search object.


I also updated Bio::Index::Blast to use SearchIO instead of BPlite to
parse the report.  For those not familiar with the object, it is useful if
you have a whole lot of Blast reports which you might imagine storing in a
single file and giving people access to the raw report, like from a
website, this object gives an easy way to basically index based on query
name so you can jump to that record in the report.

I also updated Bio::Index::Blast to use SearchIO instead of BPlite to
parse the report.  For those not familiar with the object, it is useful if
you have a whole lot of Blast reports which you might imagine storing in a
single file and giving people access to the raw report, like from a
website, this object gives an easy way to basically index based on query
name so you can jump to that record in the report.


I also fixedRoot::IO to NOT try and close IO::String filehandles as this
causes segfaults (a t/Molphy.t segfault was happening for this).

I think this is particular to a specific version of IO::String but I'm not
sure - I had the same problems with ChrisM's Data::Stag trying to close
IO::String fh's too.  Something for a perl hacker to investigate I hope.


Root::IO will also not try and close a filehandle if it is STDOUT or
STDERR now.


Finally, I backed out the changes to Bio::TreeIO classes as my first
attempt to parse newick files with labeled internal nodes didn't work
(sorry Mike)  and we'll have to give it another go later on.  This fixes
the Molphy and Tree.t test failures.

-jason

--
Jason Stajich
Duke University
jason at cgt.mc.duke.edu