From: dblack@...
Date: 2003-01-23T11:27:46+09:00
Subject: Re: Can we attack the 'not enough libraries' thing straight on?

Hi --

On Thu, 23 Jan 2003, Simon Cozens wrote:

> dblack@candle.superlink.net writes:
> > Also, I would advocate not getting embroiled in issues of naming,
> > specifically hierarchical module naming (text/soundex,
> > algorithms/sort/qsort, etc.)
>
> David, I respect you and I agree with you; but on this point I have to
> differ. As I've said several times in the past, if people don't use
> authoritative names for modules, they'll use cutesy names, and you'll
> never know that Foozilla is a module to parse SGML. I know that if I
> go to CPAN and get something called SGML::Parser then it's reasonably
> likely to be a SGML parser. Naming is important!

Would it surprise you to learn that I rather knew I'd hear from you in
response to that part of my post? :-)

The thing is, names can be mapped later, and arbitrarily many views of
a module archive can be created.  My concern is that the process of
deciding up front what things should be called, what to do about
multiple versions of modules, how to deal with badly written
"official" modules, etc., is potentially endless, and that if people
are waiting for that to happen, we'll never get the modules we need.

> Please, let's not forget lessons we learnt - even today: someone was
> asking for a Ruby Lisp interpreter. Now, what would that be called?
> Language::LISP? Scheme.rb? No, of course, it was given the eminently
> sensible name of Rouge, from which one could determine, well,
> absolutely nothing at all, except that it might have something to do
> with makeup. Naming is important!

Yes, but what if there are five Lisp interpreters?  And what if some
of them are better than others?  Same for SGML parsers, etc.  Somehow
there has to be a mechanism for people to write software that they
want to write, and for it to be distributed on something other than a
winner-take-all basis.  That, in turn, means that even if there's a
naming hierarchy, it will never be the whole story.  And that,
finally, means that there's no reason to wait to write software until
that naming hierarchy is agreed upon (whatever that even means -- I
don't even know that such a process is possible).

> The things that CPAN has that Ruby doesn't have:
>         1) sensibly named modules
>         2) reasonably comprehensive documentation
>         3) that's it. Go forth and hackify.

OK, but what modules do we need? :-)  Or are you saying that (some of)
the software in question is there but hidden under names that make
people think it's not there?


David

-- 
David Alan Black
home: dblack@candle.superlink.net
work: blackdav@shu.edu
Web:  http://pirate.shu.edu/~blackdav