David Steuber <david@...> writes:
> Is anyone else having this problem? I'm building SBCL from CVS HEAD
> using version 0.8.16.20. The build results can be found here:
As I said, both during development of the Unicode branch, and when I
reported its merge on the 2nd of this month, the
sb-md5:md5sum-sequence function is broken by design: it fails to "do
the right thing" on strings.
The right thing is probably to restrict md5sum-sequence to vectors of
octets, and to have a separate md5sum-string function with an
:external-format keyword argument, which under the hood would call
md5sum-sequence on the result of a new function with signature
string-to-octets (string &key external-format).
I don't know of the status of any efforts along these lines.