Nathan,

You're definitely right! Everything seems to be working now.

Thanks for walking me through this. 



Andrew E.

On Sat, Mar 28, 2015 at 1:56 AM, Nathan Goldbaum <nathan12343@gmail.com> wrote:


On Fri, Mar 27, 2015 at 10:41 PM, Andrew James Emerick <aje2123@columbia.edu> wrote:
I feel like I may have just tied myself into a package installation knot...

I decided to try and preempt future headaches by just installing via conda ($conda install yt). This ran fine, and I can now run "yt update" without getting the error message. So now, for the commands you asked for:

which yt   ->         /home/emerick/anaconda/bin/yt
which python >     /home/emerick/anaconda/bin/python
import yt; yt.__file__ -> /home/emerick/code/yt/yt/__init__.pyc

Now, that is somewhat baffling. The final path is the path to the original yt installation (installed from source, managed by mercurial) I was using. The weird part is that that path is not located in any environment variables that I could think of (e.g. $PATH, $PYTHONPATH, $LD_LIBRARY_PATH, etc.).

I figured there was a chance it would work now, even given the above, but when running the same code as before I get a new error (though I guess maybe I shouldn't be surprised): http://paste.yt-project.org/show/5494/  

This is actually due to a very recent change by Cameron Hummels, who totally revamped the plot callbacks recently. The annoate_line callback should take two coordinates and draw a straight line between them.  Take a look at the dev docs:


And the docstring for the callback:

 



Andrew E.


On Sat, Mar 28, 2015 at 12:40 AM, Nathan Goldbaum <nathan12343@gmail.com> wrote:


On Fri, Mar 27, 2015 at 9:28 PM, Andrew James Emerick <aje2123@columbia.edu> wrote:
Nathan,

I can do "import h5py" without issue in a python script, and checked the "from h5py import _errors" statement that seems to cause the breakage. Updating via conda didn't fix the issue. I have hdf5 and h5py installed via conda.

Doing some more checking it looks like (for whatever reason) when running "yt update" it is looking for a version of the hdf5 library that I don't have anywhere. When I did the soft linking of the libhdf5.so that I do have to the libhdf5.so.6 that it is looking for, it yelled at me saying it is trying to use the HDF5 version 1.8.4 headers, but can't find the corresponding libraries. (the most up to date version I have installed is 1.8.14).

I guess a workaround would be to get the HDF5 1.8.4 library somewhere, but I'm still bothered by this... 

I'm not sure what's going wrong here.  You shouldn't need to create a softlink and you shouldn't need to install an old version of the hdf5 library. It sort of looks like you have the remnants of an old python installation conflicting with your current anaconda installation. Unfortunately it's really hard to debug this sort of thing remotely...

Can you paste the output of:

$ which yt
$ which python
$ python -c 'import yt; yt.__file__'
 
I'm looking for yt being installed in a strange place outside of your normal python installation.

You could try the yt conda package, installable via "consta install yt", which is newer than the old pre-yt 3.1 development version you had installed.

You could also try installing yt from a fresh clone of the yt mercurial repository.  See the instructions here:


Finally, you could build a fresh conda package based on the yt mercurial repostiory.  That will require a small modification to the conda recipe to point it at the mercurial repository instead of the release tarball on pypi.

Hope you're able to sort this out.

-Nathan

PS Python packaging is literally the worst.


Thanks again.



Andrew E.


On Fri, Mar 27, 2015 at 5:18 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:


On Fri, Mar 27, 2015 at 2:15 PM, Andrew James Emerick <aje2123@columbia.edu> wrote:
Nathan,

Thanks for the response. I have been delinquent in updating yt. 

Running "yt update", however gives me this error: http://paste.yt-project.org/show/5492/

Looks like an issue with finding the libhdf5.so.6 library that I couldn't figure out on my own. I had installed yt from source as to not disturb the python, numpy, hdf5, etc. installations I already had (managed by anaconda). I have libhdf5.so.9 in my LD_LIBRARY_PATH (I already tried soft linking .so.9 as .so.6 .. understandably it threw an error).


The error is happening when python tries to import h5py.  Can you import h5py in a script that doesn't use yt?  How did you install h5py?

$ conda update -f h5py

might fix this.
 
Have you run across this before?



Andrew E.

On Fri, Mar 27, 2015 at 1:04 PM, Nathan Goldbaum <nathan12343@gmail.com> wrote:


On Fri, Mar 27, 2015 at 10:00 AM, Andrew James Emerick <aje2123@columbia.edu> wrote:
Hello everyone,

I just ran into an error in using "data_source" kwarg with the SlicePlot function (error message copied below):

slice = yt.SlicePlot(ds, axis, field, data_source=region)
File "/home/emerick/.local/lib/python2.7/site-packages/yt-3.1dev-py2.7-linux-x86_64.egg/yt/visualization/plot_window.py", line 1816, in SlicePlot
    return AxisAlignedSlicePlot(ds, normal, fields, *args, **kwargs)
TypeError: __init__() got an unexpected keyword argument 'data_source'

I have no problem doing the above with ProjectionPlot. If anyone knows what could be causing this I'd appreciate the help. Otherwise I can do some digging into the code myself.


Hi Andrew,

This should work, but might not if you haven't updated yt in a while.  Can you update to the latest development version and try again?

-Nathan
 


Thanks,
Andrew E.
--
Graduate Student 
Columbia University
Department of Astronomy

_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org



_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org




--
Graduate Student 
Columbia University
Department of Astronomy

_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org



_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org




--
Graduate Student 
Columbia University
Department of Astronomy

_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org



_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org




--
Graduate Student 
Columbia University
Department of Astronomy

_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org



_______________________________________________
yt-users mailing list
yt-users@lists.spacepope.org
http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org




--
Graduate Student 
Columbia University
Department of Astronomy