When selecting elements from the MongoDB, the query is limited to the selected variant(s) or the base variant. The base variant filter used to select only those records for which var_type == "base" or those which do not contain a var_type column at all. This led to inconsistent results of get_net for nets where some records to have a var_type column and some do not:
When the net is fetched the first time, all elements are present. Pandas will insert NaN as the variant type for rows in which var_type is not present. If the user then writes the entire net (or collection) back to the database, those NaNs will be inserted into the records that previously did not contain var_type. On subsequent calls of get_net, those records thus do not fulfill the base_variant_filter anymore (because var_type is present and not "base" but NaN). Fix this behaviour by adapting base_variant_filter to also include elements for which var_type == NaN.
When selecting elements from the MongoDB, the query is limited to the selected variant(s) or the base variant. The base variant filter used to select only those records for which
var_type == "base"or those which do not contain avar_typecolumn at all. This led to inconsistent results ofget_netfor nets where some records to have avar_typecolumn and some do not: When the net is fetched the first time, all elements are present. Pandas will insertNaNas the variant type for rows in whichvar_typeis not present. If the user then writes the entire net (or collection) back to the database, thoseNaNs will be inserted into the records that previously did not containvar_type. On subsequent calls ofget_net, those records thus do not fulfill thebase_variant_filteranymore (becausevar_typeis present and not"base"butNaN). Fix this behaviour by adaptingbase_variant_filterto also include elements for whichvar_type == NaN.