On Apr 26, 2006, at 22:45, Alex Martelli wrote:
>
> On Apr 26, 2006, at 8:19 PM, Daniel Lord wrote:
>
>> Alex,
>> Just so you and anyone who cares that our platform get fair
>> treatment know. Turns out the Mac bigot was behaving badly again.
>> GMP 4.2 doesn't compile on PPC macs either and he was blaming
>> Apple's compilers. Turns out we figured out a fix involving PIC--
>> it seems Linux PPCs have the same problem. So it was gcc problem
>> not an Apple one. I called him on and asked that he fix the site
>> to avoid further embarassment ( i did that privately) but I
>> publicly asked him to include the patches for Apple i386 and PPC
>> in the source and wondered why they hadn't been adopted. That
>> struck a nerve so he set my mail address to be moderated
>> effectively quarantining me to stifle me. I figured this would
>> happen and I didn't care--I got in a few licks and he's made a
>> fool of himself. He took down the diatribe against Apple at least.
>> In my life experience, I have found that appeasing such ignorant
>> fools drunk with power only encourages them. It was a good fight,
>> I accomplished what I wanted, and I feel no loss with the gmp
>> list--I really only want it to work.
>
> Wow, thanks for the update -- that [[expletive deleted]] maintainer
> is most definitely doing GMP no favors (nor, any favors to the
> cause of open source, either). So, I guess, eventually we'll have
> to fork GMP, or something:-(.
>
> Ah well -- so the next minor 4.2.* is expected to build well -with-
> ppc on Mac both intel and PPC...?
Well, that is if he actually follows through and checks in the
changes. No telling what he'll do since he holds the keys to the
repository and the forums--he can be a a magnaimous king or a
repressive tyrant --his choice. We'll have to wait and see. they took
down the diatribe against Apple assembler and have now a geenral
comlaint about an increase in 'broken' gcc implementations that is
platform agnostic. So IMHO, that is a big win for the Mac.
What I'll do in the mean time, is put both patches up in the
MacPython wiki so at least people 'googling' for a solution will find
them. If we cannot work with him, we'll work around him as best we can.
Jsut for posterity, here is the PPC one (thanks to Jack Howarth --
[EMAIL PROTECTED]) he is building GMP 2.4 on a G5 with Xcode
2.2.1:
Daniel,
My initial guess at the source of the problem turned out to be
correct. I had thought the use of -fno-common to provide PIC code
seemed suspect. If you rebuild gmp 4.2 on a G5 with...
./configure --enable-cxx CC='gcc -fPIC' CXX='g++ -fPIC'
...the resulting build passes make check fine. We ran into a lot of
this in ppc-linux. Any shared libs built on ppc without -fPIC are
subject to weird hard to debug crashes. I seriously doubt that
-fno-common provides the same compiler behavior as -fPIC. Often
on ppc compiles I have seen both used together.
Jack
Daniel Lord
[EMAIL PROTECTED]
---
"You will never regret getting up too early,
and you'll always regret getting up too late,
but sometimes you may regret giving up too late."
-- Mountaineer's Adage
_______________________________________________
Pythonmac-SIG maillist - Pythonmac-SIG@python.org
http://mail.python.org/mailman/listinfo/pythonmac-sig
