Feature to remember successful hash algorithm when checking file

[sjb2024]

Hi there. In version 1.4.1, you added a feature to "speed up" verification by detecting the hash algorithm from filenames.

Given that it's most common for the hashes in a file to come from a single algorithm (e.g., because the checkfile is output from singular program like sha256sum, sha512sum, or rhash itself with a single hash), a similar way to "speed up" verification in practice would be to provide an option (or perhaps even make it the default) to "learn," in each separate run of rhash -c, which algorithm has been successful for a checkfile and remember that for the remainder of the run. I don't mean that anything should persist past the run of rhash -c, just that internally rhash favor hash algorithms that have previously succeeded for a given checkfile.

I think the ideal behavior might be simply to reorder the default search order of the hashing algorithms so that the previously successful one(s) are tried first, given a hash of a particular length to be verified in a checkfile; this should do no harm and speed up verification in the most common case.

Perhaps it would also be useful to provide an option that requires hashes within a checkfile to come consistently from the same source, whatever it might be, but I'm not sure how often that would be used. Still, I would use such an option over the current default behavior (given that I have many checkfiles that (1) do not always carry useful names but (2) contain long lists of hashes, all from the same algorithm). The only alternative, if I want optimal performance, is to figure out what the hashes are myself and then supply that algorithm to rhash manually.

Thanks for considering this feature request.