# from chromatic
# on Thursday 03 July 2008 17:04:
>Any new information presented to people should be useful, and
> hopefully as immediately obviously useful as possible. What do
> download rates need to be useful?
Assuming that we could obtain download counts (which we can't), I doubt
that they would be useful in selecting a module, because they still
only indicate that someone decided to try it.
We already have is_prereq as a vote of usefulness, but that doesn't
necessarily apply to solving the end-user's problem. For that, you
need something like a darkpan+scripts "is prereq" -- is_prereq_in_wild.
Given a trustworthy, static dependency scanner which would crawl $PATH
(and perhaps some set of internal projects), I would happily set a cron
job and allow it to upload a list of in-use modules weekly. I suppose
it could even post them as a file in my cpan directory.
Others might desire more anonymity, or not have a pause account. As
long as it checks the 02-packages.details.txt to make sure it isn't
revealing internal module names.
A flat list of dependencies is a good start. Useful extra info might
include: an accurate frequency count, is it most used in modules or
scripts, and is it patched (with a publicly posted patch?)
Could it be used to make a nice report of internal and external
dependencies? That might be an incentive to get it installed --
then "would you mind anonymously posting some information about the
open-source code you use?" would be a simple opt-in.
As for presenting the information, I still think wxCPANPLUS plugins are
the only way we'll be able to avoid the centralization bottlenecks.
--Eric
--
A counterintuitive sansevieria trifasciata was once literalized
guiltily.
--Product of Artificial Intelligence
---------------------------------------------------
http://scratchcomputing.com
---------------------------------------------------
