[Bioperl-l] bioperl-db install

Keith Allen kallen@paragen.com
Wed, 13 Mar 2002 14:35:45 -0500


Hmm, yes, perhaps I'm getting a bit ahead of the curve here trying
to mix and match components before everything is set to go.  I took
your advice and got the current code out of cvs, and predictably
enough I still get a passle of warnings and failed tests.  Probably
the only part of this that might be of value right now is the following
observations:

There's a typo in the README, it says

"To set the options for the test
suite, look into t/DBTestHarness.pm and make a file called
DBTestHarness.conf"

But in fact,   DBTestHarness.pm is looking for this file:

 $self = do 'DBHarness.conf'

(unless I'm misreading the file...)

Also the test scripts end up asking for basicseqdb-mysql.sql in
the sql directory, but that didn't come down for some reason with
that cvs dump (I added it in from the web site, but my results gave
me little satisfaction).

Let me know if the make test results would be useful at this point,
or if I should just hold off a bit.

Thanks,

                -Keith


Jason Stajich wrote:

> The map, marker, ePCR stuff was all from my - now rotting - code.
>
> bioperl-db-0.1 was built for bioperl 0.7, we've improved the object
> structure since that release so the Annotation objects and the RootI
> inheritance has changed, hence lots of the msgs and errors (confusing eh?)
>
> Do this
> % cvs -d:pserver:cvs@cvs.bioperl.org:/home/repository/bioperl login
> password:
> (enter cvs)
> % cvs -d:pserver:cvs@cvs.bioperl.org:/home/repository/bioperl co bioperl-db
>
> The bioperl-db distro that works with bioperl 1.0 will be packaged up
> after 1.0 goes out and we are confident that the bugs are shaken out.  The
> CVS version is the best option we have at this point.  I can generate a
> tarball for you if aren't able to get the CVS past a firewall.
>
> -j
>
> On Wed, 13 Mar 2002, Keith Allen wrote:
>
> > Hey Guys,
> >
> > Filled with a glowing sense of optimism, I tried to build the
> > bioperl-db package (latest distro, 0.1), on Redhat 7.2, perl 5.6.0,
> > and gcc 3.0.2.  The patient did more than burp, I'm afraid, in
> > fact, I'm going to spray the output from make test (make
> > proceeded with nary a whimper).  The summary from make
> > test looks like:
> >
> > Failed Test  Status Wstat Total Fail  Failed  List of failed
> > -------------------------------------------------------------------------------
> >
> > t/ePCRdb.t      255 65280     3    2  66.67%  2-3
> > t/ensembl.t     255 65280    14    1   7.14%  14
> > t/map.t         255 65280     6    5  83.33%  2-6
> > t/marker.t      255 65280    19   18  94.74%  2-19
> > Failed 4/14 test scripts, 71.43% okay. 26/152 subtests failed, 82.89% okay.
> > make: *** [test_dynamic] Error 29
> >
> > Though it may well be that some of the errors are just in the
> > test scripts (there is a lot of carrying on about Bio::Root::RootI being
> > deprecated).
> >
> > Being a programmer of very little brain, I went ahead and did
> > the make install, and then tried to run load_seqdatabase.pl (after
> > using the supplied SQL code to build the basicseqdb).  I ended up
> > hitting a number of snags, but I'm guessing that these are not
> > informative before I know how bad the state of my build is based on
> > the failed tests.
> >
> > Any thoughts?
> >
> > -Keith Allen
> >
>
> --
> Jason Stajich
> Duke University
> jason@cgt.mc.duke.edu