CPAN Autobuild [was Re: State of the Art wrt XML/XSL.]
On Sun, Jul 16, 2000 at 02:23:37PM -0400, Michael Alan Dorman wrote:
>
> Learn to create the debs yourself, then. It's not hard---in fact, you
> can do most of them using a template where only about five lines
> change, and it only takes about five minutes a shot.
>
> Of course, some would suggest that the nigh-mechanical nature of the
> process suggests it should be automated,---to them I say that I did a
> totally clean install 5.005 on a Solaris box, and when I installed
> Bundle::CPAN using CPAN (my first action) the perl installation
> promptly rendered itself unusable because certain things didn't build
> because the Solaris box wasn't set up for it, and it didn't detect
> failures adequately, etc. I reinstalled perl and decided to deal with
> dependencies and such by hand, installing each module individually.
>
> That has _never_ happened on my Debian box, in part because _there's a
> human in the loop_. Anyone who doesn't understand the value of that
> in keeping a working system needs their head examined.
>
> Mike.
>
As I previously said, the first thing I miss and that's why I start
this thread, is a 'deb-make' like procedure that let me create
quickly the structure for a standard debian-perl package. This
way, most CPAN modules can easily be translated into a debian
conforming packages with no much change. Automation is not the
first goal but can be add using an exception mechanism permitting
to override some of the automation build for CPAN package that
need it. Adding some Debian specificities to MakeMaker could
also help to integrate those specifities upstream, making the
whole thing more easy to maintain and update (Eh! you could even
have a separate Perl for Debian SDK!!!). After that, adding
a cpan-update target into the debian/rules of perl package will
simply be a welcome goody for everyone :)
Understand me. I agree with all your comment about the necessity
to have a human behind to check any automation process. However,
we can certainly help a lot if we automate some stedious task
with the possibility to override the mechanism, "fixing" by
the same way and helping the upgrade path to next version of
Perl.
Frankly, I'm not a Perl Hacker... not even a wannabe. My first
contact with CPAN was done last week, when I have to install
the XML::XPath module. Finding it useful, I decide to make
a deb of it, which let me find about the installation process
of CPAN. Copying an arbitrary debian/directory from a libperl
package, changing the maintainer name and package name, adding
the dependencies by hand was all I need to done. A simple script
could take care of ALL of this, leaving the copyright notice,
README note, and other similar thing to the maintainer. Briefly,
a job for deb-make and consors. Missing some experiences to begin
such a project however, I decided to share this idea here and that's
the results.
Take note, I'm ready to put some time right after the OLS (next week)
if someone can help me for the design and maybe provide me some clue
(like a standard parametrizable debian/rules). I'm pretty sure this
can only be seen as a plus for Debian, adding some more perl
developpers to the party :)
So please, keep constructive and stop flaming. I want clean, simple
and global solutions. No quick hack for Debian/Perl Gods only.
I'll set a more detailed design next week (after the OLS) if no one
proposed something sufficiant to work with before.
Thanks!
--
-------------------------------* *-------------------------
Fabien Niñoles / / fabien@tzone.org
Chevalier Servant de Sa Dame / / C15D FE9E BB35 F596 127F
Veneur Gris par la Clef / / BF7D 8F1F DFC9 BCE0 9436
Chaton pour Debian / / http://www.tzone.org/~fabien
--------------------------* *------------------------------
Reply to: