# from John Peacock
# on Monday 21 May 2007 12:24 pm:
>This sort of
>'extremely white box' testing is foolhardy at best
That's certainly true if your goal is to test the module.
I can see *some* value in the extreme white box approach for testing the
toolchain and learning whether or not we've solved the bootstrapping
problem. Bugging individual authors about a test box's broken
toolchain is pretty silly though. For example, there is one tester
with a CPANPLUS which is configured in the extremely broken way of:
"no Makefile.PL found, I'll pretend I know what should be in one
and try to make anyway and then complain about missing prereqs and
nevermind Build.PL or META.yml or anything else from this century."
So, how many hoops do we want to jump through? The installer toolchain
really needs to be force-upgraded *everywhere*. Thus, the "testing
whether your module works even though my installer toolchain is broken"
really isn't all that valuable.
>My question to the list is: should I just delete the Makefile.PL
>completely, or should I break it so that Module::Build won't get
>installed in unmonitored mode?
It sounds like CPANPLUS has it backwards. Imacat's report says it looks
for Makefile.PL, then Build.PL. But, this is fundamentally
incompatible with the older (or is it just mis-configured?) behavior
where a missing Makefile.PL would cause it to "just make shit up". So,
you can't delete it and you can't not. Great.
echo die > Makefile.PL
--Eric
--
Peer's Law: The solution to the problem changes the problem.
---------------------------------------------------
http://scratchcomputing.com
---------------------------------------------------
