% sm/gpgsm --help|grep WHIRL
Hash: MD5, SHA1, RIPEMD160, SHA256, SHA384, SHA512, SHA224, WHIRLPOOL

Oh, that merely lists all has algorithms Libgcrypt supports (iirc, within some sensible range). So yes, gpgsm would support that but I have never encountered one and I guess only gpgsm would be able to create such a CMS message.

I have see that you used all kind of precautions to avoid problems with the different types (enums/defines). However in --enable-maintainer-mode -Wswitch is enabled for a reasons ;-)

In rG7c91b48f0e80266cf7491c2bb7d8aabc12362643#44312, @werner wrote:

Oh, that merely lists all has algorithms Libgcrypt supports (iirc, within some sensible range). So yes, gpgsm would support that but I have never encountered one and I guess only gpgsm would be able to create such a CMS message.

No, that is a list of algorithms that gpgsm chooses to allow:

static int
our_md_test_algo (int algo)
{
  switch (algo)
    {
    case GCRY_MD_MD5:
    case GCRY_MD_SHA1:
    case GCRY_MD_RMD160:
    case GCRY_MD_SHA224:
    case GCRY_MD_SHA256:
    case GCRY_MD_SHA384:
    case GCRY_MD_SHA512:
    case GCRY_MD_WHIRLPOOL:
      return gcry_md_test_algo (algo);
    default:
      return 1;
    }
}
...
        digests = build_list("Hash: ", gcry_md_algo_name, our_md_test_algo );

In rG7c91b48f0e80266cf7491c2bb7d8aabc12362643#44314, @werner wrote:

I have see that you used all kind of precautions to avoid problems with the different types (enums/defines). However in --enable-maintainer-mode -Wswitch is enabled for a reasons ;-)

Then I propose to change the signature of these functions so that we use a plain int for the algorithms. I implemented the policy described in "that document" and it explicitly allows the use of Whirlpool for some uses.