David: I think you hit the nail on the head. If gcc was present on the
system python was built on, /sw/lib/python2.2/config/Makefile would have a
line "CC = gcc" (instead of "CC = cc"). All extension modules that use
distutils read /sw/lib/python2.2/config/Makefile to get build parameters,
hence your problem with numeric. You didn't really need to rebuild python
though, you could have just edited /sw/lib/python2.2/config/Makefile.
-Jeff
On Sat, 22 Jun 2002, David H. Peyton wrote:
> Dear Jeff & All:
>
> Jeff's comment that, "You must have a phantom "gcc" lying around on your
> system - or at least python thinks you do," made me rethink the process,
> especially because I don't (appear to) have gcc laying about (see below).
>
> I **believe** that I had obtained my python build as a binary via "apt-get".
> Thus, might the problem be that whomever uploaded the python binary had gcc
> on her/his system?
>
> I don't know the answer to that question, but I did rebuild python, this
> time from source, and now numeric-atlas builds without a problem.
>
> By the way, I use the cvs updates to unstable for everything.
>
> The KDE releases got me going with apt-get instead of just using source and
> building on my machine. I would 'apt-get' binaries, and then update them if
> there is a newer source/unstable package available. Since I have unstable in
> my path, there is generally no available apt-get package unless it matches
> the unstable version. Thus, I don't get much this way, but the speed of
> installing the available stuff is pretty good.
>
> But there may be a downside if the system that the builder has some
> substantial differences from mine (or yours?).
>
> Experts...What do you think?
>
> Best to all!
> david
>
> On 6/22/02 2:38 PM, "David H. Peyton" <[EMAIL PROTECTED]> wrote:
>
> > It says:
> >
> > % which gcc
> > gcc: Command not found.
> >
> > -david
> >
> > On 6/22/02 2:21 PM, "Jeff Whitaker" <[EMAIL PROTECTED]> wrote:
> >
> >> You must have a phantom "gcc" lying around on your system - or at least
> >> python thinks you do. what does "which gcc" say?
> >>
> >> -Jeff
> >>
> >> On Sat, 22 Jun 2002, David H. Peyton wrote:
> >>
> >>> Dear Jeff & fink-users:
> >>>
> >>> I have been trying to do some work with pymol. This depends upon numeric,
> >>> but numeric won't build on my system. I got around it with 'sudo apt-get
> >>> install numeric'.
> >>>
> >>> But then I have now installed atlas, so I want to try using numeric-atlas.
> >>> This also fails, with:
> >>>
> >>> [rydal:~] peytond% fink install pymol
> >>> sudo /sw/bin/fink install pymol
> >>> Information about 1258 packages read in 4 seconds.
> >>>
> >>> fink needs help picking an alternative to satisfy a virtual dependency. The
> >>> candidates:
> >>>
> >>> (1) numeric: Multidim'l numerical arrays for Python, using built-in libs
> >>> (2) numeric-atlas: Multidim'l numerical arrays for Python, using atlas libs
> >>>
> >>> Pick one: [1] 2
> >>> The following additional package will be installed:
> >>> numeric-atlas
> >>> Do you want to continue? [Y/n]
> >>> rm -rf numeric-atlas-21.0-1
> >>> mkdir -p /sw/src/numeric-atlas-21.0-1
> >>> tar -xzf /sw/src/Numeric-21.0.tar.gz
> >>>
> >>> {more lines...a few seconds}
> >>>
> >>> running build_ext
> >>> building '_numpy' extension
> >>> creating build/temp.darwin-5.5-PowerMacintosh-2.2
> >>> gcc -DNDEBUG -O3 -Wall -Wstrict-prototypes -IInclude -IPackages/FFT/Include
> >>> -IPackages/RNG/Include -I/sw/include/python2.2 -c Src/_numpymodule.c -o
> >>> build/temp.darwin-5.5-PowerMacintosh-2.2/_numpymodule.o
> >>> unable to execute gcc: No such file or directory
> >>> error: command 'gcc' failed with exit status 1
> >>> ### python failed, exit code 1
> >>> Failed: compiling numeric-atlas-21.0-1 failed
> >>>
> >>> I think that the building of numeric (without atlas) might be similar, but I
> >>> don't remember...
> >>>
> >>> Thanks!
> >>>
> >>> David
> >>>
>
>
--
Jeffrey S. Whitaker Phone : (303)497-6313
Meteorologist FAX : (303)497-6449
NOAA/OAR/CDC R/CDC1 Email : [EMAIL PROTECTED]
325 Broadway Web : www.cdc.noaa.gov/~jsw
Boulder, CO, USA 80303-3328 Office : Skaggs Research Cntr 1D-124
-------------------------------------------------------
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
_______________________________________________
Fink-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-users
