Hello everybody,

The proposal has me wondered if letrec-mixed is
actually a part of the proposal (in the sense that a
SRFI-24 compliant Scheme has to provide it), or if
it is just a way to explain the semantics of
internal define-syntaxen.

I would recommend against making letrec-mixed obligatory.
The reason is that tomorrow somebody wants to add
internal SRFI-9 record types, then somebody wants to add
internal define-values, and so on and so on.

Doing all this with letrex-mixed2, letrec-mixed3, etc. gets
old very fast. We get a combinatorial explosion of
possible letrex-mixed's. Already with define-syntax, define-values
and SRFI-9 we have 2^3 = 8 possible letrec's (including the
original R5RS letrec).

So IMHO it is fine to describe the effect of internal define-syntax
in terms of some helper function, but please refrain from
requiring implementations to provide letrec-mixed.

Stephan

--
Make sure you don't miss this!
http://www.win.tue.nl/scee2002

4th International Workshop on
Scientific Computing in Electrical Engineering
23-28 June 2002,  Eindhoven,  The Netherlands

"Share and enjoy."
-----------------------------------------------------------------------
S.H.M.J. Houben                 Philips Research Laboratories Eindhoven
Building: WAY3 073              Prof. Holstlaan 4
Phone: +31 40 2743497           5656 AA  Eindhoven
                                The Netherlands
mailto:xxxxxx@natlab.research.philips.com