Server-side Dataset ViewsΒΆ


This is an experimental feature.

HDF5 views of Datasets may be stored on the server to improve query performance. To use views, first specify a path to store views in the qcfractal.config.ViewSettings.

Next, generate a view for the collection(s) of interest:

import qcfractal.interface as ptl
ds = ptl.get_collection("ReactionDataset", "S22")

# Note the server will look for views in the directory specified above,
# named {collection_id}.hdf5
view = ptl.collections.HDF5View(viewpath / f"{}.hdf5")

Finally, mark the collection as supporting views:

# Update the dataset to indicate a view is available
ds.__dict__["view_available"] = True

# Optionally, you may add a download URL for the view
ds.__dict__["view_url"] = ""