> I think the Software Development Kit (SDK) approach has more appeal:
> Define a standard bundle of extra modules that people are recommended
> to install with Perl. There was an attempt at an defining an SDK
> for Perl some time ago but it seemed to flounder.)

Yes, for many reasons.

- oftentimes there are several modules that do almost the same: choosing
  which one leads into lots of nitpicking, choosing some will hurt someone's
  feelings (either the module authors' or their proponents'), choosing all
  is bloatware

- some people don't like their modules to be included in module collections

- different people understand "SDK" differently: for some, a collection
  of uncoordinated modules with uncoordinated APIs with uncoordinated
  maintenance is not an SDK

- the previous attempts tried to span the all of CPANdom

- some application areas have too little experts; some have too many

Frankly, I don't think restarting the old SDK discussion has much merit.
It has been tried a few times in the past, and look what we have now.

What I think could work is one or both of the following:

- smaller, more focused "SDK working groups" where, say, the database
  people make their module bundles 

- by fiat: some people collect their own favourites and make *their*
  own module bundles and publish them

(If people start to do this, we need a naming convention for "SDK bundles")

> [CC'd to perl5-porters to see if there's any life in the SDK concept]

-- 
$jhi++; # http://www.iki.fi/jhi/
        # There is this special biologist word we use for 'stable'.
        # It is 'dead'. -- Jack Cohen

Reply via email to