[Toybox] FYI musl's support horizon.

[Rob Landley]

On 8/24/21 6:27 PM, enh wrote:
> yeah, i think he (and i, when i have my bionic hat on) have an easier problem
> than you ... this kind of thing isn't too hard in the c library because it's
> generally considered okay to just shrug and return -1 with errno == ENOSYS or
> whatever.
> 
> whereas as an "app", your users expect you to do the thing. (which at best tends
> to mean you have an untested/barely tested "success" case. no-one worries that
> their return -1 has bitrotted, but the non-inotify path or the non-O_TMPFILE or
> whatever ... that's a lot more likely!)

I keep telling people I could spend a focused year on JUST the test suite and
they don't believe me. When people talk about function testing vs regression
testing vs coverage testing I get confused because it's all the same thing? You
have to test every decision point (including the error paths), you have to
exercise every codepath (or why have that codepath?) and you have to KEEP doing
it because every distro upgrade is going to break something.

In my private emails somebody is trying to make the last aboriginal linux
release work and the old busybox isn't building anymore because makedev() used
to be in #include <sys/types.h> and now it's moved to <sys/sysmacros.h>. (Why? I
dunno. Third base.) And yes I confirmed that version skew using the Centos
release that other guy poked me about last week. It has the old one, my laptop
(and man7.org) has the new one, it changed somewhere in between, no idea why.

Linus is much better about avoiding this in the kernel, but you still get it
with the /sys directory because Greg KH is SUCH an asshole. And Peter Anvin is
on a mission to add gratuitous build dependencies to every project he touches
for reasons I've never understood despite repeatedly asking him to explain...

Rob