It's useful to have the metadata.json file loaded with an object, and really useful to have the <same_name>.json file loaded. But these may be cached in separate directories if one was updated but not the other in the latest zenodo version. The catalog logic that could solve this problem is currently handled in Catalog and the highest-level load function, so that sub-loaders only need to solve the simpler problem of just understanding the data format itself.
Maybe sxs.load should take the include_json argument, and pass the file path to the loaders where relevant.
It's useful to have the
metadata.json
file loaded with an object, and really useful to have the<same_name>.json
file loaded. But these may be cached in separate directories if one was updated but not the other in the latest zenodo version. The catalog logic that could solve this problem is currently handled inCatalog
and the highest-levelload
function, so that sub-loaders only need to solve the simpler problem of just understanding the data format itself.Maybe
sxs.load
should take theinclude_json
argument, and pass the file path to the loaders where relevant.