On 01/09/2025 19:09, Collin Funk wrote:
Pádraig Brady <p...@draigbrady.com> writes:
p.s. I'm thinking cksum might deprecate the explicit -a sha{224,256,384,512},
instead preferring -a sha2 -l {224,256,384,512}, which is more symmetrical and 
neater.
Though for interop compat we would not change the output tag, keeping 
SHA{224,256,384,512}.

I agree with this in principle. However, it would cause some breakage.
Off the top my head some announcements on info-...@gnu.org would have
commands that no longer work:

     $ ./build-aux/announce-gen --release-type=alpha --package-name=coreutils \
     --previous-version=1 --current-version=9.7.265-e028c65 \
     --url-directory=nowhere --cksum-checksums \
     --gpg-key-id=8CE6491AE30D7D75
     [...]
     Here are the SHA1 and SHA256 checksums:
de6ff43c89db596102a437e1e76175b3dec15570 coreutils-9.7.265-e028c65.tar.gz
       JVDvV5rSM+MFe2TxAIyaeW+1p7d/pA77seKJ6/xfk0I=  
coreutils-9.7.265-e028c65.tar.gz
       0cef5cc5a22d54a80ecc01976e48c68388757c6e  
coreutils-9.7.265-e028c65.tar.xz
       6NOdMI+ZuhaWCt5LV6Pyv1GZBwJx6IK9B0famJzeF0M=  
coreutils-9.7.265-e028c65.tar.xz
Verify the base64 SHA256 checksum with cksum -a sha256 --check
     from coreutils-9.2 or OpenBSD's cksum since 2007.

Well we'd still allow `cksum -a sha256` but only document `cksum -a sha2 -l 
256`.
Anyway I'll think more about it.

cheers,
Padraig

Reply via email to