>>>>> On Sun, 3 Jan 2010 00:51:12 -0800, Eric Wilhelm <enoba...@gmail.com> said:

 >> Working from the inside outwards, the part we need first is probably
 >> the permissions database

>From PAUSE perspective the requirement before that would be a way to
declare which language(s) an author wants to be indexed in.

Also from PAUSE perspective I think it's an advantage to have perl6
hosted by PAUSE because there is a big overlap of the developer
population.

  > Does this need anything special to support Perl 6 namespace ownership or 
  > would it just be a duplicate of the Perl 5 "schema"?

It's special in that perl 6 allows to run multiple versions of a
module together within the same process. Conflicts only arise on
namespace+version level. Indexers must keep old versions available.

  > Or, from the outside again:  What sort of indexes are required and thus 
  > what data do they need the gatekeeper to manage?

  > There was a mention (by japhb) on #toolchain about parrot modules being 
  > hostable on multiple servers, which required the client to various 
  > metadata from multiple sources.  Also:
  > 
http://blogs.perl.org/users/geoffrey_broadwell/2009/12/welcome-and-say-hello-to-my-friend-plumage.html

 >> , and then a PAUSE-like thing on top of that.

  > If we require a metafile for Perl 6 distributions, could the p6-ness 
  > potentially be auto-detected and handled by the same upload tool?

Autodetection always gives cause for trouble later on when it fails for
somebody. I'd prefer to require something explicit. But contrary to
David's believe I expect to get more mileage from having the explicit
declaration within the package. Why? (1) Tarballs will exist that have
both perl5 and perl6 versions. (2) People *will* upload into the wrong
directory.

-- 
andreas

Reply via email to