[Bioperl-l] Status 0.7
Thu, 1 Feb 2001 09:30:47 -0500 (EST)
On Wed, 31 Jan 2001, Todd Richmond wrote:
> On 1/30/01 10:28 AM, "Hilmar Lapp" <firstname.lastname@example.org> wrote:
> > In parallel to fixing known bugs (known from the bug-tracker) the package
> > shall be tested on various systems and against the projects we want to be
> > compatible with (Mac, Win32, Perl 5.004, Ensembl, bioperl-gui, bioperl-corba,
> > which will certainly reveal additional bugs/problems.
> I had a few hours to spare today, so I ran the tests for the bioperl-live
> branch on MacOS 9.1, Perl 5.004. Besides the difficulty of getting all the
> external dependencies working (which I'll write up for the documentation),
> there's a few problems that will need to be fixed.
> General issues: There's no "make test" on Macs (at least that I've found).
> When running *.t manually, every test script that hard-coded 't/filename' as
> a input file failed. When the 't/' was removed, scripts ran normally.
Todd - are you running them from the bioperl-live dir as in
perl t/SeqIO.t ??
You need to do it this way since we are looking for a relative directory.
Of course, you aren't running this cmdline probably so you can't choose
the relative directory, hmm. The filenames have to be 't/filename' for
unix so that when we run make test it can find them. I guess we can try
and be very clever here and contruct the names but the code gets ugly real
quick since we cannot rely on File::Spec to give us nice fileseparators.
Cross platform is going to be interesting...
> Index.t, SearchDist.t, Variation_IO.t all cause MacPerl to crash - both
> running the script and checking for syntax errors! The latter makes it hard
> for me to diagnose the problem...
> Chain.t, LiveSeq.t, and Mutator.t all fail because cluck in Carp is not
And you cannot install Carp from CPAN?
> OddCodes.t, SeqAnal.t, SeqStats.t, and Tools.t all seem to pass the tests
> but give an error message: Bio:Root:RootI.pm Line 118: Odd number of
> elements in hash list.
will have to look at that.
> SeqFeatureProducer.t fails - problem with // instead of :: ?
yep, not sure how we're going to deal with this.
> Sim4.t passes 9/10
> not ok 5
> # Test 5 got: '/nfs/disk21/birney/prog/wise2/example/human'
> (Cellwall:Documents:Downloads:bioperl-live:t:Sim4.t at line 31)
> # Expected: 'human'
This has to do with the fact that you are running on the mac and so the
fileseparator is '/' on unix, yet it is '::' on the Mac, so when we go to
split up the filename we are dependant on the system we are running on. I
guess the logic will have to be a little bit smarter about this or we make
the test less restrictive.
> Blast.t Passes 1..24, can't find gzip to finish tests
This is part of steve's model to support gzip, we should probably make
sure and skip these tests on non unixes.
> BPpsilite.t fails - doesnąt even get started. Spews error messages and
> ends with:
> # Cellwall:Documents:Downloads:bioperl-live:t:BPpsilite.t has too many
> DB.t all put one test skipped - can't connect to GenBank?
no LWP probably
> GDB.t failed 8/11 tests
> I'm not sure, but I don't think the LWP stuff is working...
> largefasta.t fails
> # Can't call method "seek" on unblessed reference, <GEN0> chunk 2.
> File 'MacPerl ź:site_perl:Bio:Seq:LargePrimarySeq.pm'; Line 195
largeprimaryseq is probably not going to work on a mac.
> largepseq.t failed immediately - couldn't open tempfile
tempfile creation is not going to work in the way we need it to on macs...
I'm not sure there is going to be a way to make it work on the mac unless
a machead has time to try and find fixes.
> I'll try to get these entered via the web page so that we can track them.
> Dr Todd Richmond http://cellwall.stanford.edu/todd
> Carnegie Institution email: email@example.com
> Department of Plant Biology fax: 1-650-325-6857
> 260 Panama Street phone: 1-650-325-1521 x431
> Stanford, CA 94305
> Bioperl-l mailing list
Center for Human Genetics
Duke University Medical Center