Hello,
Are you interested in helping out SBCL development, but unsure how to
get started? If so, I (Richard Kreuter) am looking for some
collaborators:
(0) to compile a good list of details that ANSI leaves unspecified,
implementation-defined, implementation-dependent, "it is an error",
etc. I have a list of the implementation-defined details here:
http://www.progn.net/static/lisp/cl-document-required.txt
This list needs work. (Note that such a list can be useful even if
it's incomplete; so closely reading just about any extent of the
CLHS will be helpful to this effort.)
(1) To write tests demonstrating what SBCL does about those details in
the list described in (0) and perhaps also the "Proposed Revisions
and ANSI Clarifications" page on Cliki. My immediate interest is to
have a database describing just what SBCL actually does around the
edges (remaining agnostic about whether our behaviors are intended
or accident, or guaranteed to users).
(For simplicity, I'd like it if such unit-tests were written using
SB-RT with as little additional rigmarole as we can get away with.)
(2) To compose english descriptions, suitable for inclusion in the
manual, of what SBCL does about those details in the list described
in (0). SBCL is strictly non-conforming for our lack of
documentation of implementation-defined details; so it might be good
to have documentation-oid writeups at hand for eventual inclusion in
the manual.
(This point and the previous are clearly related, but I think
writing tests and documentation can be done in parallel.)
For my part, I'll collate whatever we can come up with, and will make it
all publically available (though not necessarily by including them in
the canonical SBCL source tree unless other SBCL developers agree that
the work is valuable there); I'll also take care of whatever
administrivia the effort requires. In exchange for your contributions
you'll get
(a) your name in a file somewhere (in big letters, if you prefer),
(b) the thanks of at least one SBCL developer (who spends too much of
his hacking time just determining what SBCL does around the edges),
(c) in theory, eventually, maybe, a better SBCL,
(d) a number, probably equal to zero, of SBCL-themed thank you gifts.
If you think you'd like to collaborate on these goals, please contact me
at kreuter@....
Thanks,
Richard Kreuter
|