[SciPy-dev] sdist problems.
pearu at cens.ioc.ee
Sun Feb 17 12:29:15 CST 2002
I have struggled with the sdist problem and then building from the
tar-ball allmost 5 hours. There were _lots_ of bugs but a good news is
that just few minutes ago I got
setup.py build # from tar-ball
>>> import scipy
finally working. Huh.
There are still many things to clean up, though. Some of the things you
I'll commit my changes and get back to you. Ok.
On Sun, 17 Feb 2002, Prabhu Ramachandran wrote:
> >>>>> "PP" == Pearu Peterson <pearu at cens.ioc.ee> writes:
> PP> On Sun, 17 Feb 2002, Pearu Peterson wrote:
> >> Did you remove the MANIFEST file?
> PP> Sorry, I was not clear here. I mean you _should_ remove the
> PP> MANIFEST file. It is a bug that MANIFEST is in the CVS
> PP> repository. Instead we should have MANIFEST.in or no MANIFEST
> PP> at all in CVS.
> Thanks. Shouldn't we do a cvs remove MANIFEST in that case and remove
> BTW, I just updated my copy of the cvs tree and removed the manifest
> file in my local copy and now get this error:
> $ python setup.py sdist -d /tmp
> Traceback (most recent call last):
> File "setup.py", line 126, in ?
> File "setup.py", line 85, in install_package
> File "/skratch/prabhu/scipy/cvs/scipy/setup_scipy.py", line 8, in configuration
> config = default_config_dict()
> File "scipy_distutils/misc_util.py", line 247, in default_config_dict
> if full_name:
> UnboundLocalError: local variable 'full_name' referenced before assignment
> So i added a full_name="" in the function and it seems to work okay.
> I've commited this trivial change to CVS.
> The sdist seems to build fine. I havent checked to see if this really
> works properly or not but I guess it should.
> For some reason gui_thread.tests is commented out in
> Can this be uncommented or is it there for a reason?
> Also, the tutorial directory is not packaged as part of the sdist. Is
> this by design?
More information about the Scipy-dev