Closed GoogleCodeExporter closed 9 years ago
I forgot to mention, I am using gmp5.0.5, mpfr 3.1.1 and mpc 1.0.0
I have been looking at mpc 0.9 & mpc 1.0.
mpc 0.9 had a macro RNDC, which has been named MPC_RND in mpc 1.0
Could this be the problem? What can I do to resolve this? I'll keep digging and
put up an update if I find something.
Original comment by thakur.g...@gmail.com
on 8 Sep 2012 at 6:41
I think it is fixed in the SVN repository. Can you try that?
Original comment by casevh
on 8 Sep 2012 at 6:44
I have the same problem. Are you saying the zip package is outdated?
Original comment by gokoproject
on 8 Sep 2012 at 10:19
[deleted comment]
Sorry for spamming. But if I deleted the folder that was extracted by the
latest zip, and replaced it with SVN checkout, I get this after running the
setup
Traceback (most recent call last):
File "test_pack.py", line 2, in <module>
import gmpy2
ImportError: libmpc.so.3: cannot open shared object file: No such file or
directory
>> sudo locate libmpc.so
/usr/lib/libmpc.so.2
/usr/lib/libmpc.so.2.0.0
There is no libmpc3, how do you get so.3 ?
I am on Ubuntu 11.10
/opt/local$ ls
gmp-5.0.5 gmp-5.0.5.tar.xz gmpy-read-only include lib mpc-1.0.1
mpc-1.0.1.tar.gz mpfr-3.1.1 mpfr-3.1.1.tar.gz share
Original comment by gokoproject
on 8 Sep 2012 at 10:34
I've been trying to release a beta2 version but work has gotten in the way. I
haven't tested with MPC 1.0.1 but I'll do that later tonight.
You need to tell gmpy2 where the mpc and mpfr libraries are located. Try:
python setup.py build_ext -f -Ddir=/opt/local install -f
Original comment by casevh
on 8 Sep 2012 at 11:04
Using the latest svn code worked for me. However some of the mpfr tests failed,
looking at the output it seems the issues are -
1) o/p text for InvalidOperationError differs on my PC compared to that in the
test, which should be fine
2) an issue with math functions returning numbers with 15 decimal places,
compared to gmpy2 functions returning 16 decimal places, I guess this has more
to do with my python settings
3) lastly - gmpy2.local_context() returns <gmpy2.ContextManagerObject>,
instead of context(....) as expected by the test ... is this ok?
Thanks for the help!
Original comment by thakur.g...@gmail.com
on 9 Sep 2012 at 2:25
Several problems all occurred at the same. First, there was a minor change in
MPC between 0.9 -> 1.0rc1 and the final release of 1.0. That is fixed in SVN
and will be released as beta2 soon.
Second, the -Ddir option should specify the top-level directory that contains
GMP, MPFR, and MPC. It should not include a trailing /lib or /include as they
are automatically added by setup.py.
Lastly, I'm fixing the failures in the test suite. Those failures should be
fixed in beta2 also.
Thanks for the bug report. Please report any other issues you encounter.
Original comment by casevh
on 9 Sep 2012 at 5:38
gmpy2 2.0.0b2 has been released and should solve these issues.
Original comment by casevh
on 12 Oct 2012 at 6:07
Same problem with gmpy2 2.0.0b2, see issue 64
Original comment by alexande...@gmail.com
on 10 Dec 2012 at 9:11
Original issue reported on code.google.com by
thakur.g...@gmail.com
on 8 Sep 2012 at 5:37