numfmt: add --zero-terminated option #7686
Conversation
| &prefix[1..] | ||
| } else { | ||
| prefix | ||
| }; |
There was a problem hiding this comment.
This is probably just some implementation detail, but one of the GNU compatibility tests depends on this behavior. As per the guidelines, I haven't examined the GNU implementation for numfmt, but here are some manual tests I did for compatibility:
$ printf "A\nB 1001 C\x00D E\n2002 F\x00" | ./src/numfmt -z --field=3 --to=si | cat --show-all # official test
A B 1.1k C^@D E 2.1k F^@⏎
$ printf "A \nB 1001 C\x00D E\n2002 F\x00" | ./src/numfmt -z --field=3 --to=si | cat --show-all
A $
B 1.1k C^@D E 2.1k F^@⏎
$ printf "A\n B 1001 C\x00D E\n2002 F\x00" | ./src/numfmt -z --field=3 --to=si | cat --show-all
A B 1.1k C^@D E 2.1k F^@⏎
The above output is from the GNU version of numfmt, but it's identical to the uutils version output (with this patch of course).
|
GNU testsuite comparison: |
|
There is some compilation error: |
ea20122 to
81911f9
Compare
|
GNU testsuite comparison: |
@cakebaker Thanks, this should be fixed now. I'm new to this project and didn't realize there are tests within CI is passing now except for what looks like an unrelated test for Unfortunately it doesn't look like I have permission to rerun it, unless I'm missing something. So I'll have to wait until a maintainer can rerun it to verify. (I could push an empty commit, but don't want to pollute the commit log just for this.) |
|
Thanks! |
This partially fixes #1280. With this patch, the GNU compatibility tests for
numfmt --zero-terminatedpass:One thing to note is that this patch does an
unwraponString::from_utf8becauseBufRead.split(0)results inVec<u8>instead ofString. This isn't a regression though because the rest of the program already assumes input can be put intoString. Here's an example from themainbranch:I don't see a bug report for that so I'm happy to file one. I just didn't want to do more than one thing in this PR.