From: Piers Harding <piers@...>
Date: 2003-01-24T16:07:56+09:00
Subject: Re: Distribution not CPAN Re: Can we attack the 'not enough libraries' thing straight on?


I agree with you that ultimately distributions, and "certified
collections" of packages are probably an ultimate goal, but I would
suggest that one reason why this works for RedHat is that RedHat is a
commercial entity, and they (as an entity) have the man power/time and
the money to do it.  Additionally - Perl5 is a distribution - it
contains a lot of packages that people generally use such as libwww,
CGI, Data::Dumper, the list goes on for a long way.

Saying that CPAN is a tangled web of dependencies is not viewing it for
what it is.  It is a place that a community can publish its work, to a
given set of standards, knowing that that work can be resonably searched
for.  Developing a repository of modules, and surround information
should be kept separate form how we access that information - for
example the CPAN shell neatly deals with package retrieval, including
dealing with mirror site, and manages drilling into the dependencies
issue.

In fact - I urge you all to check out the many facets of CPAN, so that
you can understand it fully.
http://pause.cpan.org - authors portal
http://search.cpan.org - main directory interface, with access to
documentation, versions, and how to get in touch with authors, and
interest groups etc.
http://rt.cpan.org - a relatively new effort to assisst with
maintenance, and bug tracking.

then there are any number of supporting sites such as
http://www.perldoc.com, etc.

Cheers.


On Fri, Jan 24, 2003 at 12:18:56PM +0900, John Carter wrote:
> On Thu, 23 Jan 2003 dblack@candle.superlink.net wrote:
> 
> > We've all heard this: "Ruby is great, but it doesn't have the
> > equivalent of CPAN."
> 
> Sigh! I was going to keep out of this one, but the responses so far have
> sucked me into it....
> 
> I don't need CPAN and I don't want such a beast.
> 
> I want a distribution. A bundle.
> 
> A CPAN is a tangled web of dependencies.
> 
> A distribution (like Redhat / Debian / Mandrake / ...) is that tangled
> web snapshotted, bundled, flattened out, smoothed and co-tested.
> 
> ie. When I install version X.Y of any distribution I get...
> a) A Large bundle of goodies.
> b) Where each goodie has been tested to operate at some reasonable level
>    of stability.
> c) And the elements of the Bundle have been tested to interoperate!
> 
> Item a) means I can get stuck in and start working straight away. The
> batteries are included.
> 
> Item b) means I'm not on the bleeding edge so I have reasonable confidence
> that I can deploy the latest stable version to a customer and it will
> just work.
> 
> Item c) means the risk of mutually incompatible module versions has been
> removed for me.
> 
> You may object that distributions remove flexibility, some people wanting
> slim bundles some wanting fat.
> 
> Just the same as you can get tiny Linux distributions I envisage there
> will be tiny Ruby distributions.
> 
> Want to script something that must be deployed on a thousand Windows
> clients but @@%$&^%$ windows cmd.exe is too pathetic to
> contemplate?
> 
> Install a "tiny ruby" distro and away you go.
> 
> Want decent script engine for an embedded platform? Install a tiny linux
> distro with a tiny ruby script distro.
> 
> Want to write something large and powerful? Install the fat distro and
> away you go.
> 
> Where we are now is want module X, you must pull down item Y Version y1.y1
> and item Z Version z1.z2. (You better hope the servers for X, Y and Z are
> up when you want them.) Now you want module W as well, for that you need
> module S and module Y Version y2.y2 which is incompatible with Y version
> y1.y1
> 
> Oh dear.
> 
> I see Ruby 1.8.0 promises to have a lot more goodies bundled with it, I am
> extremely glad.
> 
> But perhaps the Ruby bundle should be small as possible. Same as the Linux
> kernel. Just Ruby and no libraries or extensions. This is so that "tiny
> ruby" distributions can easily be created.
> 
> And then there should be a Fat Ruby, developed along the lines of the
> Debian distribution, but with a far faster release cycle.
> 
> I propose that Matz strip the Ruby 1.8.0 Kernel to the minimum. Make
> it really the smallest thing that will be useful. (Perhaps a better
> definition would be the smallest thing that would not have to be patched
> to make the "Fat Ruby" distribution.)
> 
> I propose that at least two Sourceforge projects be started up.
> 
> A "tiny ruby" that can be the engine for a Windows installer / scripter.
> 
> A "Fat Ruby" that evolves very much like the Debian project, but hopefully
> faster. The "Fat Ruby" distro V1.0 should aim to have _every_ mature /
> usable item on RAA in it.
> 
> 
> 
> 
> John Carter                             Phone : (64)(3) 358 6639
> Tait Electronics                        Fax   : (64)(3) 359 4632
> PO Box 1645 Christchurch                Email : john.carter@tait.co.nz
> New Zealand
> 
> John's law :-
> 
> All advances in computing have arisen through the creation of an
> additional level of indirection, the trick is to work out which
> indirection is actually useful.