FW: First Attempt at Crypto Library

Simon Marlow simonmar@microsoft.com
Tue, 22 Apr 2003 13:00:37 +0100


=20
> This reminds me, for my ginsu project=20
> http://repetae.net/john/computer/ginsu/
> I implemented interfaces to curses and part of OpenSSL's cryptography
> routines. Would there be interest in putting those into the general
> haskell librarys as well?

As far as including more code in the libraries that we ship with GHC
goes, the gates are pretty much closed right now - if your library is
small enough it might be able to sneak in, but we're actively trying to
keep a bound on the amount of code we have to build/ship with GHC.

We're aware that this makes life difficult for those trying to supply
libraries for use with GHC, and both Simon & I think this is a big
problem.  We've suggested[1] that what the community as a whole needs is
an independently-maintained library infrastructure that supports
building and packaging libraries in a system- and compiler-independent
way.  This is a BIG undertaking, which we don't have the time to tackle.
So far no-one has stepped up to the plate (with the possible exception
of Henrik Nilsson who recently distributed some code which takes an
initial stab at the problem).

This doesn't mean that using the CVS repository and putting your
libraries under fptools/libraries is ruled out; indeed there are lots of
libraries under fptools/libraries that aren't shipped with GHC by
default (ObjectIO, GLUT, X11, HGL, HaXml, etc.).  We welcome more
libraries in the source tree, and that's one way to get access to a
build system which already does a lot of the work for you - but the
fptools build system doesn't provide the packaging facilities that we'd
want from a more general library infrastructure.

Cheers,
	Simon

[1]
http://www.haskell.org/pipermail/glasgow-haskell-users/2002-November/004
322.html