[Toybox] [PATCH] fallocate: remove uClibc workaround to benefit macOS.

[Rob Landley]

On 10/23/19 2:54 AM, scsijon wrote:
> enh, it's changed to uClib-ng. Found at: https://uclibc-ng.org/ and still
> running good. Latest is 1.0.32 (Codename Reissdorf Kölsch) released 15.10.2019.

Last time I looked at that fork I wasn't impressed:

http://lists.busybox.net/pipermail/buildroot/2016-December/180102.html

But as I pointed out a year later, there are active clones of OS/2, BeOS, and
AmigaOS so what's "dead" really?

https://lists.uclibc.org/pipermail/uclibc/2017-March/049250.html

It's been another couple years since then, maybe he actually got a unified NPTL
implementation or headers that aren't stale glibc forks. I'm not sure how it
would _matter_ in a world that has bionic and musl, but...

(Ok, I do know how it would matter: the guy maintaining the fork said he's
working on it because uClibc is the only thing that supports some ancient
hardware platforms. Um... https://www.openwall.com/lists/musl/2015/05/30/1 says
ARC and Xtensa are the ones musl should add support for to finally let uClibc
die. Both architectures survived the https://lwn.net/Articles/749292/ purge but
nobody's bothered to submit support for them to musl in the last 4 years. QEMU
has a qemu-system-xtensa but not arc? Anyway, musl's left architectures on the
table. So has llvm. That's why the old stuff survives.)

Rob

P.S. If you're really bored that original thread continued on the old uClibc
mailing list at https://lists.uclibc.org/pipermail/uclibc/2016-December/049243.html