JanusGraph
Enable query.batch by default
AFAIK, most if not all users benefit from query.batch option (https://docs.janusgraph.org/configs/configuration-reference/#query). I don't see why we shouldn't enable it by default. A lot of new users won't bother reading docs carefully or tuning their configs, and they are likely frustrated by the poor performance without query.batch.
+1 for having query.batch enabled by default. Probably this change should be targeted to 1.0.0 version of JanusGraph
I just wanted to create a PR for this so we can already include in the rc1 release candidate, but then I've noticed that we actually describe 3 different scenarios for batch processing in our docs:
- no batch processing (currently the default)
- unrestricted batch processing (would be the default if we only enable
query.batch) - limited batch processing (added by @rngcntr in #2516)
and I'm wondering whether unrestricted batch processing is really the mode that we want to enable by default. Doesn't the limited batch processing solve the problems that users might have with batch processing which is probably the reason why it has been disabled by default since early Titan days?
Unreasonably large backend queries are the only reason I can imagine for batch processing to be disabled.
+1 for query.batch=true and query.limit-batch-size=true by default.
