Bulk post to quiver-mongoapi instead of single calls?
In the current workflow, a JSON file is posted to our MongoDB each time an evaluation has taken place, i.e. when a workflow has been applied to a GT workspace and the resulting OCR-D workspace has been evaluated by dinglehopper
.
However, when Quiver stops for some reason (e.g. when there has been an error during the Nextflow processing), some runs for an OCR-D release have already been posted to the DB while the following ones are missing. This causes problems in the front end, namely in the display of the average values for a metric.
One consideration to mitigate this has been to collect all evaluation results and post them to the DB in bulk if all runs have been successful.
Edited by Michelle Weidling