[Bioperl-l] Bundle::BioPerl and Pre-reqs
Sendu Bala
bix at sendu.me.uk
Mon Oct 23 15:59:01 UTC 2006
Chris Fields wrote:
> ...
>> The current situation is: CPAN users will get all optional external
>> dependencies without using Bundle::BioPerl. Manual installers of bioperl
>> (from tar.gz, from cvs etc.) must install Bundle::BioPerl manually to
>> get full functionality.
>
> I don't think forcing is necessary, so a CPAN installation shouldn't force
> someone to install optional modules. Graph.pm, for instance has a few
> optional modules, and the tests which use those get skipped and pass so the
> installation proceeds w/o problems. We could do the same (any tests using
> those optional modules display the reason why they are skipped).
I should clarify and say that that's what happens in Bioperl as well.
The 'forcing' that I talk about is simply what I assume will happen if
the user has CPAN set to automatically install dependencies. The user
could say 'no' to every question regarding the installation of
dependencies that CPAN discovers and Bioperl would still install fine.
So really the difference between the current situation and, say, the
situation when 1.5.1 was released, is that the CPAN user doesn't have to
use Bundle::BioPerl for full functionality anymore, but can still chose
not to install all the optional external modules.
The difference is the possible default behaviour. Those users that
auto-install dependencies get all the optional ones, whereas in the past
they would not have. I have to point out the benefit of this behaviour:
those people that don't care and just want it to work are more likely to
get an installation that does just work. People who know what they're
doing can still do what they want.
Before we decide what to do I guess we need hard confirmation of how
CPAN will actually behave with the current Makefile.PL. Any ideas how we
can find out?
It would also be good to have more options to break the current tie
(Nathan is for keeping PREREQ_PM populated, Chris is for having it
empty, I can go either way)...
More information about the Bioperl-l
mailing list