[Bioperl-l] "progress": useful changes vs. "shiny new thingie"
bix at sendu.me.uk
Thu Nov 16 02:55:41 EST 2006
Sendu Bala wrote:
> Brian Osborne wrote:
>> If I'm understanding you correctly you're saying that someone checking
>> bioperl-live, HEAD, would not see a Makefile.PL if we chose to use
>> yes? And there would be a Makefile.PL in the distributions, yes?
> Yes, that is what I had meant. Though now following Hilmar's suggestion,
> I might create a stub in CVS for Makefile.PL that tells you to use
> Build.PL. I don't think it appropriate to have a working Makefile.PL in
> CVS though; not the old one for reasons outlined previously, and not one
> generated by Build.PL, because generated files don't need version
> control and only confuse matters when they become controlled.
> Ok, so the current plan is:
> # 1.5.2 keeps old Makefile.PL and gains HEAD's META.yml
# 1.5.2 gains Build.PL (and ModuleBuildBioperl.pm) so that people who
read the docs on the website/elsewhere (which will be updated to explain
the new Build.PL system) don't get confused when that doesn't work,
for wide testing, and to get people used to it for the next release. The
old Makefile.PL will have 2 changes: Brians bug-fix and a little message
pointing out that Build.PL is now the preferred installation system, but
Makefile.PL still works.
> # CVS core gets a Makefile.PL stub telling you to use Build.PL
> # CVS db, run, network stay with Makefile.PL until after final release
> of 1.5.2, then move to Build.PL with Makefile.PL stub (because they
> don't have a 1.5.2 branch)
> Is everyone happy with that, or are there further suggestions for
> something better?
More information about the Bioperl-l