od: output differs when previous files fail #15530

Activity

1. cakebaker

   added

   U - od

   on Oct 19, 2026
2. fixfon commented on Oct 31, 2026

   

   fixfon

   on Oct 31, 2026

   Contributor

   More actions

   I checked this against GNU coreutils 9.12, and I think the data difference is on the GNU side.

   Smaller repro, no JSON needed:

   mkdir d
   printf 'abcdefghijklmnopqrstuvwxyz0123456789\n' > f
   od d f
   od f
   

   GNU od 9.12 with od d f prints the first 16 bytes of f, then a line of zeros, and stops at 0000040. Running od f on its own dumps all 37 bytes and ends at 0000045. So once the directory operand fails, GNU loses the rest of the data from the next file. uutils prints the full dump of f here, the same as od f, which looks like the correct behavior to me.

   The missing final offset line is a separate issue, tracked in #15990 (#16035 is working on it).

   So I'd keep uutils as it is for the data part, and the GNU behavior is probably worth reporting upstream.
3. Devel08 commented on Oct 31, 2026

   

   Devel08

   on Oct 31, 2026

   SponsorContributorAuthor

   More actions

   2. I checked this against GNU coreutils 9.12, and I think the data difference is on the GNU side.
   Smaller repro, no JSON needed:
   mkdir d printf 'abcdefghijklmnopqrstuvwxyz0123456789\n' > f od d f od f

       GNU od 9.12 with `od d f` prints the first 16 bytes of `f`, then a line of zeros, and stops at 0000040. Running `od f` on its own dumps all 37 bytes and ends at 0000045. So once the directory operand fails, GNU loses the rest of the data from the next file. uutils prints the full dump of `f` here, the same as `od f`, which looks like the correct behavior to me.
       The missing final offset line is a separate issue, tracked in [#15990](https://github.com/uutils/coreutils/issues/15990) ([#16035](https://github.com/uutils/coreutils/pull/16035) is working on it).
       So I'd keep uutils as it is for the data part, and the GNU behavior is probably worth reporting upstream.
   

   Thank you, I'm on the same opinion, I'll report that to upstream :)
4. Devel08 commented on Oct 31, 2026

   

   Devel08

   on Oct 31, 2026

   SponsorContributorAuthor

   More actions

   It seems that Anthropic's OSS scanner found out about that before me, anyways, fixed upstream in coreutils/coreutils@171b53a
5. Devel08

   closed this as completedon Oct 31, 2026