mac os x + netcdf + pykst

Peter Barry barryp at uchicago.edu
Tue Nov 21 03:16:21 UTC 2017


Thanks! Here is a brief update on progress.

- After downloading and unzipping the binary and running the app, I get an OS version error (i’m running 10.10.5, but it requires 10.12). I’m able to bypass that by running that error by running the binary contained within the app, which appears to work. The example netCDF file is recognised and read in without issue. Is the 10.12 min os x version necessary? If not, would it be possible to reduce that to > 10.8 maybe?e.g. https://stackoverflow.com/questions/39310079/mac-app-package-backwards-compatibility-error-you-cant-use-this-version-of-t

- Getting pyKst up and running was not straightforward, but I managed to get the sine-wave example to work with dirty workaround (which is likely not sustainable and it would be wiped out upon updates). Here’s a brief timeline of what I tried - after loading pykst in ipython and attempting to establish the connection, kst opens and immediately crashes. The cause of this crash appears to be related to a mac issue - I’m not sure whether this is the correct method, but to get pykst to work, I symlinked the binary contained in the .app into /usr/local/bin so it can be run from the command line. But, it turns out I can successfully call the binary directly ("/Applications/kst2.app/Contents/MacOS/kst2”), but when i run the symlink, it crashes with “Abort trap: 6” - I have no idea about why this is happening. However, if I modify pykst to point directly to the app path, bypassing the symlink, it appears to work, and I get the sine wave output.

I’m happy to write a short set of instructions, expanding on what i’ve included here - but it’s not elegant, and I don’t know whether this is how you want this to be installed.

Thanks, Pete


On 20 Nov 2017, at 14:11, Barth Netterfield <barth.netterfield at gmail.com<mailto:barth.netterfield at gmail.com>> wrote:


Ahhh... a Mac OSX user!

We now have a nightly build for Mac OSX:  https://github.com/Kst-plot/kst-build/tree/macos

Triggered by your query, I just updated the kst web site to reflect this.

Can you try that one and let me know

i) if it works at all? (and if you do get it to work, a web-site compatible mini-blurb on how to install it? I don't use macs, so I have no idea how to install this).

ii) if it had netcdf

iii) if the newest version of pyKst works with it (and if you had to make changes, what those changes were).

I will attempt to fix anything you find here.

Thanks!

cbn

On 2017-11-19 10:45 PM, Peter Barry wrote:
Fellow kst-ers,

I am having a bit of trouble installing/configuring kst on os x, and I hope you can provide some guidance.

My goal is to use kst to read netCDF files and take advantage of the python scripting using pykst. I’m on a macbook pro running os x 10.10.5, with python 2.7 run from anaconda package with Qt5 installed. Here is what i’ve tried so far:
- Installed pykst manually, as per the instructions from https://kst-plot.kde.org/pykst/ - this appears to run in an ipython.
- Installed the kst v2.0.8 .dmg binary, and ran the first scripting example (the sine wave) from the docs, which, after some simple modifications to pykst.py, worked without issue. However, there appears to be no support for netcdf files (i.e. no netcdf reader shown in Data Sources in Help - Debug Dialog)
- Installed the v2.0.7 .dmg binary (after uninstalling v2.0.8). This version appears to contain support for netcdf, with the reader present in Data Sources, recognises netcdf format (but config is greyed out) and i’m able to open and view example files. However, when I run the python example script, after a few noSuchFun() calls, kst crashes and closes.
- Attempted to install from source, but failed. I downloaded the git repo, ran cmake, and got an error after Qt5 was found, which claimed that Qt4 was required.

Of course, the other option would be to use the dirfile format. However, the application will require writing data from 3000 sensors at a rate of ~500 Hz, and its my understanding that with dirfiles, each sensor would require its own file, and i’m not sure if 3000 open files at once is a good idea (i’m happy to switch if that assumption isn’t valid).

Any help would be appreciated. Let me know if there is any other information that would be helpful.

Thanks, Pete


--
C. Barth Netterfield
416-845-0946

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kst/attachments/20171121/f646559b/attachment.html>


More information about the Kst mailing list