Re: valgrind suppressions for current CVS Guile

[Bill Schottstaedt]

> These functions should not really need to do risky things, so maybe
> valgrind is pointing out some real bugs.  Could you investigate a bit?

"Upon further review", I think these were not Guile problems after all.

Valgrind does report a minor memory leak, which I forgot to bring in
to work -- it was in scm_strftime in stime.c, calling bdtime2c (line 641),
which calls scm_to_locale_string (line 541) to fill a field of a struct ("t")
that is inaccessible upon exiting the function. I can get the exact
messages if you need them.