Re: Sage trac ticket #8664 - Upgrade MPIR
Bill Hart
I understand we fixed the issue with the "no PIC" requirements for x86
Mac 32 bits ages ago. So no longer any need to delete all those
assembly files. They aren't (or shouldn't be) used on Mac x86 any
more. Jason, can you comment on this?
Regarding the error message itself, which is to do with PPC Mac, not
x86 Mac, this looks like it could be a build bug in MPIR. It's telling
you that an assembly instruction (contained in an MPIR assembly file
popcount.s) is being used that is not supported on all PPC macs of
that vintage. This bug likely got inherited as none of us has written
any PPC code for MPIR. You can try adding the indicated option, which
may fix it on this *particular* machine. How much do we care about
these ancient Macs? I am assuming here you are not overriding MPIR's
choice of options which is supposed to be correct by supplying it
directly with a CC and CXX (which you basically can't do with MPIR
safely unless you know precisely which options to provide...) If so
then I recommend building MPIR on its own on this machine and see what
is passed to CC and CXX (about line 8 of the configure output), and
have Sage pass that. You can look in the file configure.in to see the
different options that are passed on different machines and how they
are differentiated. If it does prove to be a bug in MPIR, please
report to the mpir-devel list so we can follow up and make necessary
changes.
Regarding --enable-fat, this does not work on PPC (and never will). It
is only valid on x86 and x86_64.
Re comment 68:
I would ignore the locale thing. The machine is so ancient that there
is unlikely anything we can do about it. Unless Sage commonly uses
commas instead of decimal points to represent floating point numbers
in strings passed to MPIR, it isn't going to matter in the slightest.
It looks like someone set the test to not fail but simply be skipped
because it is essentially harmless. In my opinion this has more to do
with finance than mathematics, where Euros are written 128,00 instead
of 128.00 in some countries, etc.
Bill.
On 6 November 2010 10:17, Mitesh Patel <qed...@gmail.com> wrote:
> Hi Bill,
>
> When you have a chance, could you look at the ticket to upgrade the Sage
> MPIR spkg:
>
> http://trac.sagemath.org/sage_trac/ticket/8664
>
> In particular, there are two recent comments by Leif Leonhardy:
>
> http://trac.sagemath.org/sage_trac/ticket/8664#comment:48
> http://trac.sagemath.org/sage_trac/ticket/8664#comment:68
>
> Thanks!
>
> Sincerely,
> Mitesh
>
>
Jason
> Regarding comment 48:
>
> I understand we fixed the issue with the "no PIC" requirements for x86
> Mac 32 bits ages ago. So no longer any need to delete all those
> assembly files. They aren't (or shouldn't be) used on Mac x86 any
> more. Jason, can you comment on this?
>
Yeah , It should recognize it as an "applenopic" and only use code that is
non-pic.
ie from configure
# 32bit apple darwin doesn't like our PIC format asm code
case $host in
core2-apple-darwin* | penryn-apple-darwin*)
path="x86/applenopic/core2 x86/applenopic" ;;
prescott-apple-darwin* | pentium4-apple-darwin*)
path="x86/applenopic" ;;
pentium3-apple-darwin* | pentium2-apple-darwin*)
path="x86/applenopic" ;;
i686-apple-darwin* | pentiumpro-apple-darwin*)
path="x86/applenopic" ;;
core-apple-darwin*)
path="x86/applenopic" ;;
*) ;;
esac
Maybe your machine is not one of the above triples ? , if so , I can add it in
, easy peasy
> Regarding the error message itself, which is to do with PPC Mac, not
> x86 Mac, this looks like it could be a build bug in MPIR. It's telling
> you that an assembly instruction (contained in an MPIR assembly file
> popcount.s) is being used that is not supported on all PPC macs of
> that vintage. This bug likely got inherited as none of us has written
> any PPC code for MPIR.
Yep , as far as I can remember , we have not changed any PPC code ,