Re: The difficulty or ease of packaging Perl/Python/Ruby/PHP applications
Andreas,
On Jan 23, 2008 3:07 AM, Andreas J. Koenig
<andreas.koenig.7os6VVqR@franz.ak.mind.de> wrote:
> ajk> So please be more specific, give me a few examples that have bothered
> ajk> you and maybe we can straighten something out.
Adding to the problems you've already found. I had to struggle a lot
to find an index that could mimic /dist pages. And none really does.
Firstly, the developer releases aren't indexed. Then, there's the
problem of different criterias for indexing (distribution vs module
names). For each index I could find, there was always some exception.
Sorry that I cannot be more specific as I really don't recall all the
attempts I've made.
Today, we're using /modules/02packages.details.txt.gz to get a list of
distributions with their current versions. That doesn't include
experimental releases, of course.
To get lists of distributions to mimic by-author and by-modules, we
had to resort to ls-lR, as the other indexes had subtle differences
too.
There was also the problem of tracking author changes and also
different modules with the same name (there were few of these, I can't
remember now).
I know I'm not helping much, sorry. But what I know is that if there
were a database of distributions, with known (or latest)
versions+authors+metadata ((un)?official, developer release or not,
etc) it would be way easier to track cpan.
Thanks for fixing the problems you've found. Notes:
> libcrypt-rijndael-perl 1.05.01 > 1.05
>
> An underscore in the version number has a special meaning on CPAN: it
> means this is a developer release not yet ready for production. I
> suppose the above debian package is based on
> BDFOY/Crypt-Rijndael-1.05_01.tar.gz.
> libdate-simple-perl 3.03.03 > 3.02
The problem here is that this particular version is not indexed, so we
use the unstable release, but the QA tool doesn't know about it.
> libnet-imap-simple-perl 1.17 > 1.14
>
> Fixed by re-running the indexer. I cannot recreate the circumstances
> under which the indexer failed sometime in 2006. Sorry for that, and
> thanks for bringing it to my knowledge.
Thanks for taking the time to make things work smoother for us :)
--
Martín Ferrari
Reply to: