On Fri, 30 Nov 2007 03:57:58 -0700, David Green wrote:
> Part of a solution is search.cpan.org -- if you can figure out which 
> of the 870 XML modules will be useful to you.  Another part is asking 
> on newsgroups or lists -- if you can figure out which of the 870 
> opinions offered is knowledgeable.  I think things like the CPAN 
> ratings and reviews will become increasingly important.  Of course, 
> this is all a community issue (rather than a technical issue), and 
> questions about handling reputation are certainly not limited to Perl 
> or CPAN.
> Maybe some kind of "Advisory Board" would help, where people (who 
> might be experts in various ways) can offer informed recommendations 
> on what modules make a good fit for what circumstances.  Ultimately, 
> if this is something we want, somebody needs to volunteer to organise 
> something.  (Or volunteer to figure out exactly what it is that would 
> need organising....)

I do feel strongly that we need some sort of solution to this so that Perl
6 is not merely an outstanding framework that leaves all domain-specific
extensions to the end user.

Please note that I am not arguing for inclusion in the "core"; presumably
I *am* arguing for some sort of standard flavors of P6 that are named and
supported.  If we can't solve that any better than Luke's assessment I
fear we will have sold Perl 6 short to a large community.

Part of the major attraction of Perl 4/5 was knowing how much was
core/standard; you could write programs that did everything from DNS calls
to shared memory access to database access and know that anyone anywhere
with Perl could run them.  But nowadays you need a slew of modules to
write good programs.  It would be a shame if we perpetuated in P6 the
syndrome of having to be in the echo chamber [1] before you knew what
those modules were.  We ought to be able to spread that knowledge around a
bit better.  I'd just hate to see the same situation of "For good O-O, use
Class::Accessor," "No, use Class::Struct," "That's ancient, use
Class::Std," "No, the new standard is Object::InsideOut," "That's so 2006.
All the best programmers are using Moose now."  

Okay, so we will have standard O-O in P6 so that scenario doesn't apply,
but substitute CGI/DBI/XML/etc/etc and you have a picture that will.  Does
everyone who wants to know what to use to do those things properly in P6
have to be subscribed to TPR/perlmonks/clpm/perlbuzz/use.perl.org/arrgh?

Can we find a way to make and maintain some recommendations in a way that
people can find them easily from P6 itself?  If I'm shopping for a car and
I find a place that sells a fantastic drivetrain and says that they leave
the choice of body, wheels, and seats to me I'm going to look somewhere
else because I don't have the time to research auto component integration
even though if I did I could end up with a car to die for.

Maybe what we need is an editorial team.  Developers tend more to want to
include everything, like a Slashdot page where you have to wade through
acres of dross to find something useful, because sitting in judgment on
someone else's submission is distasteful.  But editors are used to making
judgments and dealing with the consequences.  If we could find people who
would do that job perhaps they could define a few standard bundles that
end users and distro maintainers could choose if they don't want the core
alone.  Yes, it involves value judgments and we don't like to make
those and people will argue about their decisions no matter what, but does
that have to stop us?


Peter Scott

Reply via email to