Sven Panne wrote:
Ross Paterson wrote:
[...] Ideally, the new package would be sold to the MTL users as a replacement
for the old one, which would then disappear.
We could keep the old one for backwards compatibility, see below.
[...] When the new package is ready for use,
What's missing?
it is hard to tell. nobody really uses the "new" monadic library except for me
(as far as i know). the basic transformers work i think.
for some of the more exotic transformers it is not quite clear to me
how to combine them with each other -- some just require more thinking
others i don't think can be combined in such a way that the laws one might
excpect for the operations to hold. i still have some definitions
that kind of make sense, but i am not sure if those should be in the
library. in any case these are i think corner cases, that were not
in the old library anyways.
i don't like a lot of the names used in the
library, but i didn't want to change most of them, for compatability reasons.
there are also a few dubious combinators, and backward compatability
functions that would be nice to remove.
my preference on the matter would be to not go for compatability with the old library.
if your code uses the old library it can simply use the old library. when you start work on a
new program you should be able to pick which of the two libraries to use.
the most important current problem is that the library stopped working with the current cvs version of hugs, due to a type checker bug. hopefully this will simply go away when the typechecker is fixed though.
-iavor
a next step might be dropping the Unstable prefix from the module names of
the monads package [...]
Definitively. :-)
[...] But the major part would be the negotiations with MTL users. [...]
Well, it's a bit hard to negotiate with an unknown user base. I propose to
make the new monad package the default (= auto package) in the next GHC release,
but at the same time keeping the old mtl package (which would be non-auto then)
for people who still want it. Or are there technical problems with this approach?
Hmmm, Hugs has no package concept...
Cheers, S.
--
==================================================
| Iavor S. Diatchki, Ph.D. student | | Department of Computer Science and Engineering |
| School of OGI at OHSU |
| http://www.cse.ogi.edu/~diatchki |
==================================================
_______________________________________________ Cvs-libraries mailing list [EMAIL PROTECTED] http://www.haskell.org/mailman/listinfo/cvs-libraries
