[Bioperl-l] various problems with mdk bioperl package

[Guillaume Rousse]

Hello. I'm the maintainer for mdk bioperl packages. Here are some 
problems I had with latest release.

First, one test fail as part of the whole rpm build process:
t/SeqFeatCollection..........FAILED test 425
         Failed 1/432 tests, 99.77% okay

However, I'm unable to reproduce it manually issuing a 'make test' 
command in the build directory.
perl -Iblib t/SeqFeatCollection.t -> OK

Using BIOPERLDEBUG to have more verbose output, other tests fails:
not ok 421
# Test 421 got: '0' (t/SeqFeatCollection.t at line 156 fail #406)
#     Expected: '6'
not ok 423
# Test 423 got: '0' (t/SeqFeatCollection.t at line 156 fail #408)
#     Expected: '4'

This is perl 5.8.6, without thread support, on mandrake cooker. There is 
no special environment variable used during package building that could 
explain the different results.

Second, man page generation is disabled by default, using some strange 
construction in Makefile.PL:
sub MY::manifypods {
   my $self = shift;
   #print STDERR "In manifypods moment\n";
   if( 1 ) {
     return "\nmanifypods : pure_all\n\t$self->{NOECHO}\$(NOOP)\n"
   }
   else {
     return $self->SUPER::manifypods(@_);
   }
}
If the goal is just to avoid man page generation, why not
INSTALLMAN3DIR => undef ?

Third, I'd like to split the package a little bit, to avoid drawing so 
much dependencies. Here is the whole list of external dependencies of 
the current package, as automatically computed by rpm:
perl(CGI)
perl(CGI::Carp)
perl(Cache::FileCache)
perl(Carp)
perl(Class::AutoClass)
perl(Clone)
perl(DBI)
perl(DB_File)
perl(Data::Dumper)
perl(Data::Stag)
perl(Data::Stag::XMLWriter)
perl(Digest::MD5)
perl(Dumpvalue)
perl(English)
perl(Error)
perl(Exporter)
perl(Fcntl)
perl(File::Basename)
perl(File::Path)
perl(File::Spec)
perl(File::Temp)
perl(FileHandle)
perl(GD)
perl(Getopt::Long)
perl(Getopt::Std)
perl(HTML::Entities)
perl(HTML::HeadParser)
perl(HTML::Parser)
perl(HTTP::Request::Common)
perl(HTTP::Response)
perl(IO::File)
perl(IO::Handle)
perl(IO::Socket)
perl(IO::String)
perl(LWP)
perl(LWP::Simple)
perl(LWP::UserAgent)
perl(Math::BigFloat)
perl(POSIX)
perl(Pod::Usage)
perl(SOAP::Lite)
perl(SVG::Graph)
perl(SVG::Graph::Data)
perl(SVG::Graph::Data::Node)
perl(SVG::Graph::Data::Tree)
perl(Storable)
perl(Symbol)
perl(TestInterface)
perl(Text::Shellwords)
perl(Text::Wrap)
perl(Tie::Handle)
perl(Tie::RefHash)
perl(Tree::DAG_Node)
perl(UNIVERSAL)
perl(URI)
perl(URI::Escape)
perl(XML::DOM)
perl(XML::DOM::XPath)
perl(XML::Handler::Subs)
perl(XML::Parser)
perl(XML::Parser::PerlSAX)
perl(XML::SAX)
perl(XML::SAX::Base)
perl(XML::SAX::Writer)
perl(XML::Twig)
perl(XML::Writer) >= 0.4

Just having the Bio::DB branch in a subpackage would be enough to avoid 
a mandatory dependency on Ace. What else could I split ?

Fourth, they are still two scripts in the main bioperl archive relying 
on bioperl-run: bp_pairwise_kaks.pl and bp_blast2tree.pl. They should 
really be moved there, to avoid circular dependencies.
-- 
The engine falls out of the car the day after the warranty expires
		-- Murphy's Driving Laws n°18