There are very few cases where the filter test can accurately be bypassed, and generally it has to be applied to determine ACI's as well.
The idea of a filter test bypass actually violates the principle correct-simple-fast, where we should always assert that our content matches the provided filter, given the fact that partial candidates and more can exist.
In fact, a great deal of logic around ALLIDS and flagging these would be able to be removed if this was assumed to always run. It would also make consistent behaviour with regard to application of ACI and correct query execution behaviour.
I think this should default to OFF.
After discussion with @lkrispen this may be better as an investigation into the correctness and operation of grok-filter as today it may be too broad or zealous. @lkrispen has pointed out there may be a large performance improvement on large sets in this situation by avoiding the filter test.
Metadata Update from @firstyear:
- Custom field origin adjusted to None
- Custom field reviewstatus adjusted to None
Metadata Update from @mreynolds:
- Issue set to the milestone: 1.4.2
to comment on this ticket.