Looks good to me

• Revert "ctld: Only use strtoul when str is not NULL"

Hits coverity id: 1474355 in our coverity

• Don't free line buffer in getline loop

• pass bool into prange in change

• Only end ed input stream if we aren't issuing a delete

• Make delete bool

I don't see any functional changes and rrs blocking review has landed, so this looks fine to land to me.

With just misho's patch from bz

• Remove extra format specifier

• Update field sizes to match tcpcb
• Update format string

In D34421#782565, @pstef wrote:

There doesn't seem to be any specification of the diff3 behavior, so I spent some time looking for some test cases. GNU doesn't provide much directly in the source (https://git.savannah.gnu.org/cgit/diffutils.git/plain/tests/diff3) and also the mailing list isn't rich in corner cases relating to diff3. I've found some test cases for git's implementation (https://github.com/git/git/search?l=Shell&q=diff3) but I only skimmed that so I don't know how useful they are.

I am not sure if FreeBSD is upstream here, it might be harti's own private tree, it isn't 100% clear to me.

Yeah, wrong diff point with arc diff, I am sure arcanist wasn't this much trouble to use before

Add back the code changes

Update based on man page feedback

In D34421#780268, @pauamma_gundo.com wrote:

From the summary: "Add tests, first using the original provided 1.txt, 2.txt, 3.txt, but also add
lao tzu tao tests from the gnu diff3 how to guide and finally some manually
written tests which excised other cases."

Unless you really mean "precisely or surgically cut off with a sharp knife" (literally or figuratively), that should be "exercised".

Since (per https://reviews.freebsd.org/D34411#change-nX1rdgK113Nt) mg was explicitely put in the public domain, you could use it instead of lao tzu and tao.

• Add tests for ignore case
• Correct path passed to diffit when using Nflag or Pflag

• Add tests for finding function names

Update correctly

fix a return issue and add tests

This seems to fix the panic I was hitting. Before I could reliably panic by doing buildworld with -j 16 in a bhyve guest with 16 cores.

I didn't read the entire comment, correct english is
"fd did not allow write"
"fd allowed write"

I actually thought I had abandoned this, If I remember gnu chroot looks up the user before and after the chroot so it can use the correct user id. I was told when creating the diff that this is an awful idea and I should use other tools.