On 06/09/2025 19:43, Collin Funk wrote:
Hi Pádraig,
What do you think of the proposed patch?
Arguably the setting of errno belongs in Gnulib. But I still think it is
wrong to have DIGEST_STREAM return an error, but continue the program if
errno is not set. That seems like it is prone to printing uninitialized
memory. ENOMEM seems like a reasonable fallback.
Won't the lower level functions have already set that?
One might be able the verify with (ulimit -v ##### & cksum ...)
If EVP_DigestInit_ex etc. didn't set errno to ENOMEM,
then I would think it more appropriate to do that in gnulib's sha3.c
rather than in coreutils.
cheers,
Padraig.
