In D38481#882216, @imp wrote:

What do programs that call it on main do?

By the same logic, should we exclude GENERIC for all but amd64 from universe?

committed

Final pre-commit check

Fix test declarations.

Tentatively fix an issue with the definition of YYSTYPE.

Parametrize MAXBSIZE, make sure we really do cross a buffer boundary, and explain what we're doing and why.

lgtm, we should get this in now so we have a few months to see what breaks before 14.0.

Nit: alphabetize envars.

Nit: minimize the test file.

Move the check into tarfs_lookup_path().

Add a few more test and fix the whitespace test.

In D38496#877219, @lwhsu wrote:

In D38496#877177, @des wrote:

In D38496#876169, @lwhsu wrote:

I also have some WIP tests for wc(1) to be added in D35877

It's unfortunate that this wasn't approved back in the day. Does it cover anything that my tests don't?

It's my fault that haven't completed it yet but I do plan to work on it with the submitter this week. Do you mind if I go ahead to remove the duplicated ones in D35877 and merge it first? Or you want to get this differential in first? I proposed to add tests first because it would be a good practice for code refactoring, but I can wait if you want.

Add descriptions to test cases.

In D38496#876169, @lwhsu wrote:

I also have some WIP tests for wc(1) to be added in D35877

Use printf(1) instead of echo.
Add a test case with only blank lines.
Add a test case with only whitespace.

*sigh* I keep finding more stuff to fix

Fix exit codes.

Fix multiline_repeated test.

Remove unintended test file.

Add tests.

Avoid needless repetitions of a hardcoded buffer size.

Switch mbrtowc() cases around to avoid awkward line wrapping.

Avoid repeatedly checking for a null filename.

Superseded by D38496.

Now that we've restored tzsetwall(3) in stable/13, we can think about removing it from main.

Widening the scope as per discussion on IRC, will resubmit.

In D38463#875225, @kib wrote:

BTW, are there malloc()s calls in the code where sizes are controlled by the archive content? If yes (and from my casual reading there are) it is a significant issue.

Fix size check, for real this time.

Instead of hardcoding knowledge of tmp/obj-tools into depend-cleanup.sh, run it twice.

Remove unnecessary line continuations.

I'm not 100% sure hardcoding tmp/obj-tools here is the best approach; perhaps we should simply run depend-cleanup.sh with OBJTOP=${WORLDTMP} before building bootstrap tools?

See D38445.

Note that I chose to reimplement tzsetwall(3) as it was easy enough to do, but I'm not sure this is the right approach as it locks us into maintaining a patch in contributed code.

ping @cperciva as this is a regression in stable/13 which needs to be fixed before 13.2.

@val_packett.cool can you file a PR instead and abandon this review?