Diffpy_cmi-1.0 on Mac

[YONG-IL KIM]

Hi guys,

I’m trying to install diffpy-cmi on my mac with EI Capitan.

1) sudo port select --set ipython ipython27

Selecting 'ipython27' for 'ipython' failed: The specified version 'ipython27' is not valid.

2) sudo port select --set python python27

Selecting 'python27' for 'python' succeeded. 'python27' is now active.

3) and then the installation procedure was completed (cmi-install--build.txt).

4) When running the test (./runtests.sh), the test run was failing (cmi-runtests.txt). 

5) So in order to confirm the installation, after I tried to reinstall (./install –build) and to do the test again.

But the result was same. What did I wrong?

 

Can anyone who knows more than I do help me?

 

Cheers,

YONG-IL KIM.

[Pavol Juhas]

On Sun, Dec 13, 2015 at 03:45:25AM -0800, YONG-IL KIM wrote:
...

> I’m trying to install diffpy-cmi on my mac with EI Capitan.
>
> 1) sudo port select --set ipython ipython27
>
> Selecting 'ipython27' for 'ipython' failed: The specified version
> 'ipython27' is not valid.

Hi Yong-Il, the name of the ipython package in macports has changed
to py27-ipython, so that line should be:

sudo port select --set ipython py27-ipython

You can also check what packages can be selected to stand for default
ipython using:

sudo port select --list ipython

> 4) When running the test (./runtests.sh), the test run was failing
> (cmi-runtests.txt).

The segmentation fault in pyobjcryst is due to a bug which
started showing with later versions of c++ compiler. The other
failure is due to small changes in Python language. To fix these,
rebuild the CMI again with the latest sources from the online code
repositories. This is a bit tricky, because there were quite a few
updates in the meantime and sometimes the C++ compilation fails
when there are old build files mixed up with latest sources.
To avoid this, do

./install --update
scons -C src/libobjcryst --clean
scons -C src/pyobjcryst --clean
scons -C src/libdiffpy --clean
scons -C src/diffpy.srreal --clean

Here the sources are at their latest versions and any generated files
have been cleaned. After "./install --build" the runtests.sh script
should pass without complaints.

Hope this helps,

Pavol

PS: If running scons...--clean is too tricky, another option is to
start anew. Extract the source distribution again and run

./install --update
./install --build

--
Dr. Pavol Juhas
Condensed Matter Physics and Materials Science Department
Brookhaven National Laboratory
P.O. Box 5000
Upton, NY 11973-5000

[YONG-IL KIM]

Hi, Dr. Pavol Juhas,

Thank you for your kind help. However, unfortunately I could not finish installing diffpy_cmi-1.0 on Mac although I followed your comments.

As you can see, there are some errors in #5 libObjCryst.

 

# 5 libObjCryst --------------------------------------------------------------

scons: Reading SConscript files ...

scons: done reading SConscript files.
scons: Building targets ...
scons: *** [build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.os] Source `build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.cpp' not found, needed by target `build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.os'. scons: building terminated because of errors.


 

If you can spend your time on helping me fix it, please.

 

Regards,

Yong-Il Kim.

2015년 12월 16일 수요일 오전 5시 31분 29초 UTC+9, YONG-IL KIM 님의 말:

[Pavol Juhas]

On Wed, Dec 16, 2015 at 7:29 AM, YONG-IL KIM <yiki...@gmail.com> wrote:

...

# 5 libObjCryst --------------------------------------------------------------

scons: Reading SConscript files ...

scons: done reading SConscript files.
scons: Building targets ...
scons: *** [build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.os] Source `build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.cpp' not found, needed by target `build/fast-x86_64/cctbx/cctbx/eltbx/sasaki_tables_01_12.os'. scons: building terminated because of errors.


 

If you can spend your time on helping me fix it, please.

Hi Yong-Il,

This is still an error due to a lingering build files.  The sasaki_tables_01_12.cpp has been removed

from the library, but somehow the scons build system still expects it to be there.  I would recommend to remove

the whole "build" directory in the libobjcryst source directory and then resume the install-build from step 8:

rm -rf src/libobjcryst/build

./install --build=5-10

Hope this helps,

Pavol

[YONG-IL KIM]

Hi, Dr. Pavol Juhas,

Thanks to your help, I could install diffpy-cmi-1.0 on my Mac.

However, when running an example (calcpdfc60), I have some errors.

Please refer to the attachments.

Regards,

Yong-Il Kim.

2015년 12월 16일 수요일 오전 5시 31분 29초 UTC+9, YONG-IL KIM 님의 말:

Hi guys,

2015년 12월 16일 수요일 오전 5시 31분 29초 UTC+9, YONG-IL KIM 님의 말:

Hi guys,

[Pavol Juhas]

On Sat, Dec 19, 2015 at 05:04:18AM -0800, YONG-IL KIM wrote:
> Hi, Dr. Pavol Juhas,
> Thanks to your help, I could install diffpy-cmi-1.0 on my Mac.
> However, when running an example (calcpdfc60), I have some errors.
> Please refer to the attachments.

ipython --pylab
...
ValueError: unknown locale: UTF-8

Hi Yong-Il, This is unrelated to DiffPy-CMI. There is some error
in your matplotlib installation or in your locale (language,
date format, etc.) setup. Please check the value of your LANG
environment variable. In my case it is "en_US.UTF-8", but if set to
"UTF-8" the "ipython --pylab" would give the same ValueError.
I hope this helps, if not, please send further questions
about this to either matplotlib or locale mailing list.

Thank you,

Pavol

[YONG-IL KIM]

Hi, Dr. Pavol Juhas,

I could completely solve the problem (UTF-8) by adding the text 

to .bashrc as follows:

export LANG=en_US.UTF-8

export LC_ALL=en_US.UTF-8

  

Thank you very much for your kindness again.

Regards,

Yong-Il Kim.

2015년 12월 19일 토요일 오후 10시 38분 7초 UTC+9, Pavol Juhas 님의 말: