Closed yymao closed 4 years ago
Note: This change does not affect catalog users. User-facing API will remain the same. User experience may be improved due to the potential speed improvement during catalog loading.
@wmwv already has script for this, but see also https://github.com/LSSTDESC/gcr-catalogs/issues/351
Newer DC2 DPDD catalogs (source, forced source, dia source) are now all in parquet format with consistent schema across files, and they all use readers that are subclasses of
DC2DMCatalog
.On the other hand, the original DC2 DPDD object catalogs are still in hdf5 format, and hence for now we need to maintain two readers that are similar in nature. The reader of the DC2 DPDD object catalogs also has to fill in missing columns.
We should convert the DC2 DPDD object catalogs into the same parquet format as the newer DC2 DPDD catalogs.