Conversation
Codecov Report
@@ Coverage Diff @@
## master #249 +/- ##
==========================================
- Coverage 74.11% 73.86% -0.25%
==========================================
Files 13 14 +1
Lines 2503 2525 +22
Branches 594 609 +15
==========================================
+ Hits 1855 1865 +10
- Misses 428 435 +7
- Partials 220 225 +5
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
Thanks @dschwoerer ! I do agree that a generic solution here would be better, if it can be done. One thing that can be done is the unit conversion and data labels. xhermes and xstorm use attributes: e.g. Those attributes could be in the BOUT.dmp files, enabling automatic conversion |
|
I think that would make most of the pluggins work unnecessary 🎉 Do we want to say "SI" - or are we happy with eV as temperature? All I would be missing is The other thing that may be needed is if the log of a quantity is evolved. I think STORM is doing that ... |
|
I would say that as long as the units are recognised by UDUNITS or perhaps pint then it should be ok -- That might also extend to log units too, but I suspect that might be trickier or just not exist in any existing convention -- so you'd either need to just document it somewhere, define new units, or perhaps even just only write out the non-log form? |
Alternative to xHERMES :-)
It allows to use xBOUT for loading any BOUT++ simulation. That way I don't have to think what simulation has generated the data. I just load it, and as long as an appropriate extension is installed, e.g. https://github.com/dschwoerer/xbout.modules.hermes for hermes (as a proof of concept).
After all, I don't want to do something hermes specific. I just want to have decent units ...
If there are some post-processing that would only ever a STORM user do, sure.
But I rather feel the other way round. I would like to have the same interface for my EMC3 simulations and my BOUT++ simulations. But the ones are
ds.emc3.*the othersds.bout.*- I want to avoid this to go worse ...Note that there is a pip bug with submodules and editable installs: pypa/pip#11467