On Mon, 21 May 2007, John Peacock wrote:
I received a complaint against my SVN::Notify::Build distro:
https://rt.cpan.org/Public/Bug/Display.html?id=27183
that has me annoyed. I provide a compatibility Makefile.PL so that users
without Module::Build installed will get bootstrapped into the modern era.
For whatever reason, this wouldn't work with the OP's system and caused his
smokes to just burn CPU. That's not my problem.
My problem is that the OP seems to think I should provide a fully EU::MM
compatible Makefile.PL, because installing M::B is too much of a hardship. I
can see that the OP has entered another bug:
https://rt.cpan.org/Public/Bug/Display.html?id=27208
where Dave Rolsky responded much as I did.
Yeah, this was a very frustrating exchange.
I _still_ am not sure I understand the core problem.
To say it in my own words, this person does not want to "pre-install"
Module::Build on the systems she (I think) is using to do CPAN testing.
However, since CPANPLUS does ship its own (out of date?) version of
Module::Build, things mostly work, because if a module only has a
Build.PL, CPANPLUS will use its own version of M::B.
However, that CPANPLUS-specific version of M::B is not visible in the
normal load path, so our compatibility Makefile.PL code fails to see it
when trying to find M::B::Compat.
It seems that her argument is that she should expect to be able to have
things "just work" with the CPANPLUS-specific M::B. That is problematic
for what I hope are obvious reasons. This is a totally non-standard
install of a module, that may be out of date, that is not easily
upgradeable, and may contain custom code of an unknown nature (unknown to
us module authors out in the "wild").
If CPANPLUS had not provided this bundled M::B, I think she'd be forced to
install Module::Build as a prereq to testing, much as she'd have to
install CPAN(PLUS) and EU::MM. Asking her to also install M::B hardly
seems unreasonable.
This is especially true since to the best of my knowledge, what was done
with CPANPLUS was very much a nasty hack undertaken by one person to solve
a perceived problem which is probably better solved in another way.
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? Or should I just mark it as WONTFIX and change nothing? Please don't
spare my feelings if you think I'm being too much of a hardass about this...
I think she should just freaking install Module::Build on the testing
machines. To say that this is now a prereq for doing testing is hardly a
ridiculous statement, any more than suggesting that you should have a
vaguely recent version of EU::MM.
-dave
/*===================================================
VegGuide.Org www.BookIRead.com
Your guide to all that's veg. My book blog
===================================================*/
