[Aboriginal] And a new version of busybox joins the pile...

[Rob Landley]

I've been swamped at work (contract ends on the 31st and my todo list is
longer than that, wound up called into work on Saturday for crisis du jour).

But I've also been trying to get a release out with uClibc 0.9.32 and
Linux 3.0.  And I got pretty close: in the -hg version sh4 is fixed, and
the mips thing I spent last weekend tracking down was PEBKAC on my end.
My todo list is down to two targets, only one of which I care that
deeply about (Powerpc threading is still broken, segfaults on launch).
I think I'm ok letting m68k have a regression for one release, I can fix
that later.  (I need to try to get QEMU to run m68k stuff again.  Once
that happens it becomes much higher priority.)

Well, BusyBox 1.19.0 just got added to that new package list, and it
breaks stuff.

The first thing it broke was adding ubi_tools.c, which breaks with the
Linux 3.0 headers as I complained about here:

  http://lkml.indiana.edu/hypermail/linux/kernel/1108.1/02863.html

Easy thing to do is revert that patch.  Except the SECOND thing the new
busybox breaks is that patch can't create files.  Which is my fault,
commit 8027a202a89b where I was asked to fix something unrelated and
broke this along the way.

Don't ask me why between November 2010 and today, NOBODY NOTICED BUSYBOX
PATCH CAN'T CREATE FILES.  I mean... dude?  Does anybody actually test
this stuff?  (My excuse is I hadn't updated from 0.18.3 to 0.18.4 since
it's presumably a bugfix only release for bugs I hadn't hit.  The whole
"quarterly releases of new development versions with bugs in between"
thing seems to have fallen by the wayside...  Oh well.)

Anyway, still trying to get a release out ASAP.  But I'm hip deep in
patch.c diagnostics again, and the last time I looked at this code was
apparently November 2010...

Rob