Previously, np.concatenate was used to aggregate all row runs in a chunk.
This meant that we needed twice us much memory - one set for all row runs and one for the concatenate result. Now, the result is preallocated, and pyrap.tables.table.getcolnp is used to dump data
directly into the result array.
Previously, np.concatenate was used to aggregate all row runs in a chunk. This meant that we needed twice us much memory - one set for all row runs and one for the concatenate result. Now, the result is preallocated, and
pyrap.tables.table.getcolnpis used to dump data directly into the result array.