[SciPy-user] -lf77compat failure - SciPy on AMD64 Sun Studio
Mon Jul 14 22:21:53 CDT 2008
Robert Kern wrote:
> On Mon, Jul 14, 2008 at 21:35, Michael Abshoff
> <email@example.com> wrote:
>> Hope this helps. If anybody cares I can send my notes on the build
>> (including Python) to the list so that someone interested can stuff them
>> in the wiki.
> Yes, please! However, I don't think anyone who hasn't built numpyor
> scipy on Solaris will have the motivation (or background, probably) to
> pretty them up for the wiki, so they'll probably get lost if you don't
> do it yourself. But toss 'em over anyways. Actually, if you have notes
> on building anything in the Sage stack on any platform, I'm personally
> interested in seeing them.
Ok, besides the 64 bit OSX there is little that is working and of
interest to you guys. The main issue is still getting the ctypes
extension to build at all. For various reasons what is in-tree in python
is crap and lacking 64 bit Darwin support completely and I have not had
time to come up with a clean patch for ffcall (or is it fficall?).
For 64 bit OSX Scipy there was one failure with one extension that was
compiled in 32 bit mode even on 64 bit. I would assume someone is
cleverly clearing the CFLAGS or CXXFLAGS and then shooting me in the
foot, but maybe I myself was dumb and overlooked something.
Solaris in 32 and 64 bit mode is mostly done in regards to python, numpy
and scipy. I promised Stefan to run a build bot on Solaris and I have
started to put together eight builds, i.e.
a) Sparc, Sparc 64, x86 and x86-64
b) gcc with gfortran or cc with SunPerflib
So those are eight build bots in total. But progress there has also been
hampered by me working on other Sage issues.
What will hopefully happen next months is a 64 bit ATLAS+Netlib.org
Lapack+numpy+scipy on Vista 64bit, but we will see how that goes. If I
have trouble with ATLAS I will probably use the Intel MKL for now to get
up and running on the 64 bit native Sage on Windows port.
>> There seems to be no mention of Solaris on
>> I also send some build info on 64 bit OSX out a while ago and somebody
>> mentioned that those might also be well placed in the wiki.
> To this list? I'll have to go back through the archives.
It was sent to the numpy list IIRC. If they don't turn up in the
archives I can send you a copy off list.
More information about the SciPy-user