Release notes

Changes from v0.2.0 to v0.2.1

  • Improved performance of pandas and xarray loading from cdf.

  • Added nrecords_limit option to viresclient.SwarmRequest.get_between() to override the default maximum number of records in each request. Use this if a request is failing with a server error that the maximum allowable number of records has been exceeded - but this means that there is probably duplicate data on the server (old and new versions), so check the data that gets returned:

    data = request.get_between(start_time, end_time, nrecords_limit=3456000)
    ds = data.as_xarray()
    # Identify negative time jumps
    np.where(np.diff(ds["Timestamp"]).astype(float) < 0)
    # e.g [2519945, 5284745, 5481414]
    for i in [2519945, 5284745, 5481414]:
        print(ds.isel(Timestamp=i))
    # Length of day should be 86400
    ds.sel(Timestamp='2014-02-02')
    
  • Added tmpdir option to viresclient.SwarmRequest.get_between() to override the default temporary file directory. The default is selected automatically according to https://docs.python.org/3/library/tempfile.html#tempfile.mkstemp (usually /tmp). This may not be suitable when fetching large amounts of data as some machines may have limited space available in /tmp or there may be a higher performance or preferred location.

Changes from v0.1.0 to v0.2.0

  • Now use SwarmRequest instead of ClientRequest.
  • kwarg subsample changed to sampling_step.
  • Added references to .available_collections() and .available_models().
  • User credentials are automatically stored in a configuration file ~/.viresclient.ini.
  • Downloads are streamed to temporary files instead of being held in memory.
  • Any size request is now supported. Large requests are automatically chunked up.
  • Added download progress bar indicating size in MB.
  • xarray added as a dependency and .as_xarray() method added.