Re: hppa, powerpcspe: call for klibc/mksh testers
On Fri, May 18, 2012 at 10:45 AM, Thorsten Glaser <email@example.com> wrote:
> no mailing list for powerpcspe? What’s the progress of your
> architectures… I don’t see much happening on debian-ports.org,
> and taking your architectures off that infrastructure hurts
> others (I’ve “revived” m68k and kept it running on d-p.o which
> helps a bit).
> Anyway: a new klibc (2.0~rc5-1) has been uploaded, and an mksh
> companion version (40.9.20120518-1) depending on it and trying
> to use klibc for /bin/mksh-static, complete with testsuite that
> is run by the buildd (unless DEB_BUILD_OPTS contains nocheck).
> Please, please build and test these klibc and mksh versions on
> as new a sid environment as you can get (I could probably ask
> for access to an hppa lenny chroot, but… ugh!) and see whether
> it’ll work for you – klibc on sh4, for example, was broken and
> nobody noticed it until I ssh’d to a porterbox (there seem to
> not exist any for hppa and powerpcspe) to fix an FTBFS, wanted
> to test the result and saw SIGILL (then downloaded the last
> version that had built, ran it and saw a SIGILL, so _that_ was
> no regression).
So I'm afraid I don't have powerpcspe hardware anymore; I left Boeing
to go work for Google in January. Unfortunately I believe the guys I
was working with there are all really crazy overworked right now, so I
don't actually know how much progress they are likely to make on it in
the next few _years_.
Additionally, even FreeScale appears to be abandoning the e500v2
hardware; all of their new chipset families are e500mc (32-bit) or
e5500 (64-bit), both of which return to classic PowerPC FPU and can
run normal "powerpc" Debian.
So if nobody is maintaining it anymore, perhaps the best thing to do
really is to kill it off quickly?
I wish things had turned out better for that, but on the other hand I
also wish I hadn't had to sink quite so much of my life into working
around FreeScale's braindead FPU design too, so...