use Perl

• Needs some caching? (Score:1)

  by Aristotle (5147)

  Any query page is very slow to load, even after querying the same module a second time right away.
  • Re: (Score:2)

    by drhyde (1683)

    Added some caching now, which makes it quite a bit better. It's still slow though, because:

    • It's on the cheapest hosting provider I could find;
    • It uses CGI

    It's intended mainly as a proof-of-concept. Hopefully someone will pick it up and integrate it into the cpan-testers site and do all the boring work of making it mod_perl-safe :-)
• Very cool (Score:1)

  by jrockway (7171)

  This is quite helpful. I know Angerwhale is a pain to install, but now I can see that Crypt::OpenPGP is the flakiest module I depend on (and mostly because it depends on other flaky modules).
  
  Crypt::OpenPGP will be optional Real Soon Now.
  
  And BTW, I think Plagger and Jifty both have more dependencies. :)
  • Re: (Score:1)

    by sartak (7900)

    Jifty does have more deps, but I'm not convinced that's a bad thing! :)
• Circs (Score:1)

  by kwilliams (1467)

  That's pretty rad. Is there anything that will detect & show circular dependencies?

   -Ken
• some ideas that come to mind (Score:2)

  by jhi (318)

  Pretty & cool!
  
  Compute this once per night/weekend for all the modules in the CPAN to draw pretty graphs.
  
  Add an option of ignoring modules that begin with Test:: and are included only because of modules beginning with Test::
  
  
  • Re: (Score:2)

    by drhyde (1683)

    If I were to pre-compute results, then probably the thing to do is to drive it off the CPAN recent uploads list. I won't ignore Test::* dependencies, because if someone lists one of 'em as a pre-requisite then its failures are just as important as failures in any other dependency. Which, by a rather convoluted route, brings us back to the mailing list thread that spawned this, in which people were grumbling about Test::Pod etc causing failures.
• Just as efficient as the module authors... (Score:1)

  by janus (3624)

  Speaking from a porters point of view i've seen _so_ many modules with wrong and/or hidden dependencies that i hardly believe the results will be satisfactory to all of us.
  Even if you just focus on build/run-depends you'll notice many inconsistencies :-/
  
  Sure it's a nice to have - but unfortunately not as reliable as we might hope.
  
  The most effective way... would be correct dependencies or collecting them manually... or better: allow to mark wrong dependencies so we have a way to identify them and file proper
• This is marvellous (Score:1)

  by grinder (1100)

  This is a very nifty thing you have built.

  One can look at barbie's statistics on modules that fail http://perl.grango.org/wdists.html [grango.org] and http://perl.grango.org/wpcent.html [grango.org] and then rebound off domm's Kwalitee pages that provides the crucial list of which modules/dists depend on a flakey module.

  This would then allow people to hone in on unloved, critical modules and see if the author is no longer interested in them and take over maintenance (or maybe the author doesn't realise these tools exist these da
  • Re: (Score:2)

    by drhyde (1683)

    Are you signed up for the CPAN Testers BOF?
    • Re: (Score:1)

      by grinder (1100)

      Natürlich :)
• Net::SSH::Perl FTW! (Score:2)

  by djberg96 (2603)

  If memory serves, the "record" (in my experience) for most dependencies, when built with ssh1 and ssh2 support and all the optional encryption stuff, was Net::SSH::Perl [cantrell.org.uk].

  That only shows 18 because it can't figure out all the dependencies. The final tally was over 30, iirc. And that was only if you were lucky enough to get Math::Pari to build.
• Update ... (Score:2)

  by drhyde (1683)

  I've abstracted out the bit that finds dependencies, it's now on the CPAN, as CPAN::FindDependencies. It also installs a 'cpandeps' script.