[Haskell-cafe] Re: Thinking about what's missing in our library coverage

[Magnus Therning]

On Wed, Aug 5, 2009 at 8:59 AM, Bulat
Ziganshin<bulat.ziganshin at gmail.com> wrote:
> Hello Magnus,
>
> Wednesday, August 5, 2009, 11:37:23 AM, you wrote:
>
>> I don't know of any other way either.  I just strongly oppose the idea
>> that HP should take on the role of providing C lib bindings just
>> because on some platforms it's hard to satisfy the C dependencies.
>
> those some platfroms are 97% of all dowanloads and success on these
> platforms is the key to overall Haskell success. moreover, asd i
> understand the situation, lack of package manager on Windows was main
> motivation to establish HP - for unicies it's not really required

80% of all internet-related statistics are made dubious ;-)

I strongly doubt the "97% of all downloads" statement.  However,
that's not really what we are discussing here.  This is the statement
on the Haskell Platform page:

"The Haskell Platform is a blessed library and tool suite for Haskell
distilled from Hackage, along with installers for a wide variety of
machines. The contents of the platform is specified here: Haskell:
Batteries Included.

"The platform saves you the task of picking and choosing the best
Haskell libraries and tools to use for a task. Distro maintainers that
support the Haskell Platform can be confident they're fully supporting
Haskell as the developers intend it. Developers targetting the
platform can be confident they have a trusted base of code to work
with."

The way _I_ read it, HP is a set of libraries that form a supplement
to a Haskell compiler/interpreter.  Developers can feel confident
writing code against this set of libraries and it's the goal to make
HP available on as many platforms as possible.

I don't think that establishing HP was mainly motivated by the lack of
a package manager for windows, I also don't think that HP is un-needed
on Unices.  AFAIU the motivation was to 1) separate the
compiler/interpreter (especially GHC) from "base libraries", 2) to
clearly communicate what Haskell packages a developer can expect to
find on a "Haskell system", and 3) to provide users/developers with an
easy route to setting up a "Haskell system" on different OSs.

Difficulty to build a C dependency of a Haskell library should _not_
be a criterion used to decide whether the Haskell library goes into HP
or not.

Cabal is great for source distribution, but apparently there's a need
for a binary packager, especially for Windows.

/M

-- 
Magnus Therning                        (OpenPGP: 0xAB4DFBA4)
magnus＠therning．org          Jabber: magnus＠therning．org
http://therning.org/magnus         identi.ca|twitter: magthe