hashsum does not accept --length

[oech3]

$ hashsum --b2sum -l 0 /dev/null
786a02f742015903c6c6fd852552d272912f4740e15847618a86e217f71f5419d25e1031afee585313896444934eb04b903a685b1448b755d56f701afe9be2ce  /dev/null
$ hashsum --b2sum -l 8 /dev/null
786a02f742015903c6c6fd852552d272912f4740e15847618a86e217f71f5419d25e1031afee585313896444934eb04b903a685b1448b755d56f701afe9be2ce  /dev/null
$ cksum -a blake2b -l 0 /dev/null
BLAKE2b (/dev/null) = 786a02f742015903c6c6fd852552d272912f4740e15847618a86e217f71f5419d25e1031afee585313896444934eb04b903a685b1448b755d56f701afe9be2ce
$ cksum -a blake2b -l 8 /dev/null
BLAKE2b-8 (/dev/null) = 2e

Solution: Just drop the duplicated interface.

[Dolphindalt]

GNU Coreutils:

1. has a standalone b2sum command with a -l option for digest length.
2. has a cksum -a blake2b as a unified interface that also supports the -l option.
3. does NOT have a hashsum command

"hashsum" => {
  phf_map.entry(krate, format!("({krate}::uumain, {krate}::uu_app_custom)"));
  
  let map_value = format!("({krate}::uumain, {krate}::uu_app_common)");
  phf_map.entry("md5sum", map_value.clone());
  phf_map.entry("sha1sum", map_value.clone());
  phf_map.entry("sha224sum", map_value.clone());
  phf_map.entry("sha256sum", map_value.clone());
  phf_map.entry("sha384sum", map_value.clone());
  phf_map.entry("sha512sum", map_value.clone());
  phf_map.entry("b2sum", map_value.clone());
}

hashsum is a shared utility that uses the binary name it was invoked as to determine the behavior. This is the (third) duplicated interface the issue author has suggested dropping.

So what ought to be done? Two choices:

1. Remove the duplicated interface entirely
2. Fix the duplicated interface by accepting the length argument

What is the preferred approach? I don't understand why the hashsum command is incorporated into the core-utils rewrite and could use some additional context. Otherwise, this looks like duplicating an existing interface, is not GNU core-utils compatible, and adds unnecessary maintainability overhead.

[oech3]

2 people are already working on hashsum deletion. But we have many barriers we should fix at 1st.

[oech3]

I'm currently working on the ln- s cksum md5sum approach, but it is too coxplex due to clap's limitation.
If clap supported unilateral overrides_with, we could easily discard hashsum...

[oech3]

@Dolphindalt Do you know any reasonambe API for #9992 (comment) ?