New parameter added to Shaper: disable_endpoint_cache. Default value, False.
By default, if sheXer is told to consume triples from an endpoint, it will make some SPARQL queries and store the results in a local graph. If this parameter is set to True, sheXer won't save that content locally.
PRO: you'll save memory. This allow to deal with huge graphs using less resources.
AGAINSTS: decrease performance. Each query will be executed twice, as sheXer needs to iterate over the content 2 times. Also, there could be scenarios in which some triples are processed twice, as not merging the content in a local graph do not ensure that the triples processed stream-like do not contain repeated elements.