metadata for non-diagnostic output is incomplete

[rabernat]

We read metadata from available_diagnostics.log. However, "non-diagnostic" output files (e.g. instantaneous state and time-average files output according to dumpFreq and taveFreq) are not described in available_diagnostics.log. The solution has been to hard-code this metadata in variables.py: https://github.com/xgcm/xmitgcm/blob/master/xmitgcm/variables.py

But variables.py is not complete: it is missing output related to EXF, seaice, and many other packages.

The solution is very simple: someone needs to make a pull request adding the missing metadata. The format for the metadata is straightforward and can be inferred from the other entries in variables.py.

This is related to xgcm/xgcm#41. cc: @jklymak

[rabernat]

this includes PTRACER output

[AlexeyPechnikov]

I added new option "diagnostics" to resolve this and some other issues (2D/3D variables with the same names mixed in one data directory). With this option we can use custom version of the default_diagnostics.log where all the variables described:

ds = open_mdsdataset('data',grid_dir='grid',
                     prefix=['prefix'], geometry='llc',
                     diagnostics='custom_diagnostics.log')

[rabernat]

Could you please submit a pull request for this feature?

Sent from my iPhone

On Jun 25, 2017, at 2:45 AM, mobigroup notifications@github.com wrote:

I added new option "diagnostics" to resolve this and some other issues (2D/3D variables with the same names mixed in one data directory): mds_store.py.zip

With this option we can use custom version of the default_diagnostics.log where all the variables described:

ds = open_mdsdataset('data',grid_dir='grid', prefix=['prefix'], geometry='llc', diagnostics='custom_diagnostics.log') — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[rabernat]

@mobigroup: just following up. Would still love to get your PR.