On Wed, 2010-04-14 at 21:51 +0200, Andreas Rottmann wrote: > I've now started with splitting out stuff from spells, and stumbled upon > one case I'd like to get feedback on: in the foof-loop package, I'd like > to add `nested-foof-loop', a library built on foof-loop that provides > for a more concise way of expressing nested loops. That library needs (a > subset of) stream.scm[0], which I've extended in obvious ways with a few > procedures having the same interface as SRFI 41 (Streams). Now > streams.scm is very minimal, and the extensions have precedence in SRFI > 41 -- does it make sense to include them in the "ported" collection, or > should I refrain from doing so? Arguments in either direction would be > appriciated! > > [ Why not use SRFI-41 directly? streams.scm builds upon SRFI 45 > (Primitives for Expressing Iterative Lazy Algorithms), and the space > leak explained in that SRFI is essential to be fixed for > nested-foof-loop. ] > > [0] http://mumble.net/~campbell/scheme/stream.scm
If this port of nested-foof-loop can be made to produce streams which SRFI 41 can work with, I think that should be done, so that other code dealing with the streams will import SRFI 41 in order to work with them. Either way, if you use your customized version of [0], it should be a library named (ported foof-loop nested streams), because it's made for this particular port of nested-foof-loop and any other name would be pretending it's not. If SRFI 41 can't be used to work with the streams, users should import (ported foof-loop nested streams) to do so, because they're dealing with streams unique to nested-foof-loop. Sorry if I'm misunderstanding something. I'm not very familiar with these libraries. P.S. This is the type of thing which should be discussed in the Ported project's issue tracker. Let's focus on getting that setup, instead of further usurping ikarus-users. -- : Derick ----------------------------------------------------------------
