[Bioperl-l] "progress": useful changes vs. "shiny new thingie"

[Sendu Bala]

Lincoln Stein wrote:
> Sendu Bala wrote:
>     This is assuredly all valid. But I feel this is a case of a necessary
>     last minute addition. Necessary because I don't feel the old Makefile.PL
>     is up to the job. Its merely unfortunate that I only discovered this so
>     late into the RC cycle. Evidently I and all the other testers don't test
>     the installation side of things, only the test scripts in t/
>
> I tested script installation last week and it was all fine as far as I 
> could tell. Are there any details about how installation was failing?

Maybe this aspect got broken since after you tested (I don't think so 
though, judging by the fix), but scripts only got put in scripts_temp 
and nowhere else.

The major prompt for the change was needing a complete META.yml for CPAN...


>     So to clarify, is anyone actually uncomfortable or 'resigned' to the
>     idea? Does anyone feel strongly in favour of keeping Makefile.PL?
> 
> I'm sorry we have to push the release back. I was unaware that 
> Makefile.PL was broken -- my preference would have been to fix it rather 
> than to rewrite things from scratch.

... I'd have had to do far more rewriting things from scratch to get a 
complete META.yml out of Makefile.PL. I didn't want to reinvent the 
wheel, and Module::Build has lots of other benefits as well.

I'd already done major adjustments to Makefile.PL previously; the move 
to Build.PL can be seen as just one more major behind-the-scenes 
adjustment, since Makefile.PL will still exist in distributions (having 
been generated by Build.PL) and people can still install the same way 
they have done in the past.

The only real change for most users is that they now need Module::Build 
installed.