Unconfigured Smokers

[Barbie]

Hi folks,

I just spotted this [1] for one of my distributions, and wondered
whether there is any mileage in the smoker bots checking whether
CPAN/CPANPLUS has been suitably configured before being run.

[1]
http://www.nntp.perl.org/group/perl.cpan.testers/2009/12/msg6531704.html

In this case Acme::CPANAuthors::Utils provides the check in
_cpan_file(), so I can potentially catch this in tests and skip as
appropriate, but could it affect other distributions?

[2]
http://cpansearch.perl.org/src/ISHIGAKI/Acme-CPANAuthors-0.09/lib/Acme/CPANAuthors/Utils.pm

Cheers,
Barbie.
--
Birmingham Perl Mongers <http://birmingham.pm.org>
Memoirs Of A Roadie <http://barbie.missbarbell.co.uk>
CPAN Testers Blog <http://blog.cpantesters.org>
YAPC Conference Surveys <http://yapc-surveys.org>

[David Golden]

On Tue, Dec 22, 2009 at 2:52 PM, Barbie <bar...@missbarbell.co.uk> wrote:
> I just spotted this [1] for one of my distributions, and wondered
> whether there is any mileage in the smoker bots checking whether
> CPAN/CPANPLUS has been suitably configured before being run.

In my smokers, I routinely sabotage CPANPLUS configuration so that
stupid, older Module::Install distributions don't try to recursively
call it and install things.

My opinion is that no distribution should assume that CPAN or CPANPLUS
are configured.

David

[Chris 'BinGOs' Williams]

Hilarious because I do the exact opposite and routinely sabotage CPAN.pm
configuration so that Module::Build::Compat generated Makefile.PLs
wouldn't be able to install Module::Build. Well, that was the original
intention, before I started installing M::B on smokers.

It does stop other people who have CPAN invocation in their Makefile.PL
from installing stuff as well.

>:)

--
Chris Williams
aka BinGOs
PGP ID 0x4658671F
http://www.gumbynet.org.uk
==========================

[Barbie]

On Tue, Dec 22, 2009 at 10:36:30PM -0500, David Golden wrote:

I'm in two minds about this. From one perspective, yes distributions
shouldn't assume anything. However, it does seem reasonable that a basic
configuration should have been done, as that would have been done by a
user as soon as they started up a cpan/plus shell.

In my case, the parent distro expects a 02packages.txt file to be
available, which if a user had started a cpan/plus shell would be there.

If other distributions expect similar basic configuration, then it could
lead to a lot of boilerplate being added, just to protect authors
against misleading smoke reports. It just seemed like a good idea for
smoke bots to check first and throw a warning to the tester if
appropriate.

Any road up, I'm looking to patch a dependency to catch this for future
use.

[David Golden]

On Wed, Dec 23, 2009 at 2:39 AM, Barbie <bar...@missbarbell.co.uk> wrote:
> I'm in two minds about this. From one perspective, yes distributions
> shouldn't assume anything. However, it does seem reasonable that a basic
> configuration should have been done, as that would have been done by a
> user as soon as they started up a cpan/plus shell.

The problem is that they can't assume which one is configured.

> In my case, the parent distro expects a 02packages.txt file to be
> available, which if a user had started a cpan/plus shell would be there.

Then you need to check both cpan and cpanplus. That might not be true
as there is no strict need for that file if CPAN::Index is running.
It could have been purged after the Index DB is created. Or, the file
could have been grabbed direct from a minicpan and never moved to the
.cpan directory. Moreover, you can't assume a particular directory,
you actually need to check the CPAN/PLUS config files to find where
temporary files like that are stored.

All of which is to say -- if you need 02packages.txt, then you should
probably download it yourself.

David

[Barbie]

On Wed, Dec 23, 2009 at 08:44:08AM -0500, David Golden wrote:
>
> All of which is to say -- if you need 02packages.txt, then you should
> probably download it yourself.

Which is the patch I'm going to be adding to RT :)