[Toybox] [PATCH] mkpathat: return EEXIST only if MKPATHAT_MKLAST is given

[Rob Landley]

On 2/8/21 2:00 AM, Yi-yo Chiang wrote:
> I was hesitant to attach a testcase demonstrated by "cpio", as I don't think
> cpio is the real cause here. It just happens so cpio were calling mkpath in the
> correct order.

It doesn't have to be the culprit, it just has to be "this is a corner case that
didn't used to work and now it does". It could be the command, lib, libc,
kernel, weird compiler flag, alignment issue on specific embedded hardware...

I sometimes use git annotate and look at the commit to see WHY a specific test
showed up.

> If only we had a test case that directly test lib/lib.c itself?

There's a toys/example directory with various demo_*.c commands that exercise
bits of library functionality directly, and a "make test_demo_number" target.
(Elliott is picky about human_readable(). :)

Rob