This is the mail archive of the
guile@sourceware.cygnus.com
mailing list for the Guile project.
binary-io, opposable-thumb, pack/unpack (was Re: binary-io (wasRe: rfc 2045 base64 encoding/decoding module))
not being more than a scheme programmer wannabe, i don't think i can
address the lack of certain low-level features w/in guile itself (at
least not yet).
in the opinions of those of the list, is it possible to make useful
progress on something like binary-io at only the scheme level?
if not, is there anyone who will take up the task of adding any
necessary low-level features? i'd offer my help, but i don't know how
much use i would be at this point.
my understanding of the discussion surrounding binary-io so far is:
-chars are currently 8-bit in guile, but that is an accident of
history -- someday they may have a different size. so in the
future, will chars not have a fixed size? btw, is anyone working
on multibyte character support?
-ports are defined in terms of chars, and chars might not be
fixed in width to 8-bits in the future.
-if a port-centric interface is ok to use for binary-io, what happens
when one can no longer depend on the size of a char being 8 bits
in guile? would one create and manipulate a soft-port that returns 8-bit
quantities for its version of read-char?
-if a port-centric interface is not ok, what should i use instead?
concerning non-serial access to data, i thought i saw reference to
work someone was doing on a data structure that acts a bit like an
emacs buffer. for some applications, couldn't that be used? i guess
it might depend on the implementation of "buffers".
also, i still don't see how at a scheme level, how one can detect
what the native format is for an int or a long. the original design
of binary-io mentioned being able to manipulate native quantities...