On Wed, Oct 22, 2008 at 11:07:50PM +0200, Slaven Rezic wrote:

> I don't think so. This here is one of possibly many errors at the
> distance. It's not clear why this one is special to be workarounded by
> either CPAN::Reporter, CPAN, or MakeMaker. Maybe it's enough if the
> author checks for a ~/_Inline directory in the Makefile.PL and exits
> with a warning message.

My understanding is that the directory is used by Inline to cache
compiled snippets of code between runs, so that the very long startup
time needed to run compilers and so on doesn't hit the user every time
he uses an Inline-ish module.  It is therefore entirely legitimate for
the directory to exist and have stuff in it.  If a module can't cope
with that, then that module is buggy.

My suggestion of setting the PERL_INLINE_DIRECTORY variable was for the
completely different case where a tester is using two different versions
of perl, which then makes things get confused because Inline dumps their
cached binaries in the same place.  That's what the RT bug report we
looked at was all about.  I don't think that CPAN::Reporter, or CPAN, or
MakeMaker, or CPANPLUS should do anything about it.  I think that
*testers* should do something about it.  They should do something like
this:

#!/bin/sh
export PERL_INLINE_DIRECTORY=$HOME/.Inline-5.6.2
for i in $*; do
    /home/david/cpantesting/perl-5.6.2/bin/perl -MCPAN -e "install qw($i)"
done

Now, back to the directory already existing.  A module that can't cope
with it already existing is buggy, we can all agree.  However, I would
suggest that if it can't cope with the directory being left in an
inconsistent state, we can't really blame the author, and shouldn't send
him a bug report - after all, if there *is* a bug, it's a bug in Inline,
not in the module that we are ultimately testing.  So, to minimise the
risk of sending bug reports to the wrong person, I'm going to
periodically delete .Inline* for exactly the same reason that I delete
.cpanplus and a few other droppings that badly-behaved modules leave in
my home directory.

And Slaven - please don't CC me on mailing list messages.  It means that
sometimes the copy I see won't have Reply-To properly set, which is why
I originally replied just to you.

-- 
David Cantrell | Godless Liberal Elitist

   When a man is tired of London, he is tired of life
      -- Samuel Johnson

Reply via email to