To better test against problems like #5743, we should add a toy testing implementation that deals with small integer keys only (and perhaps use []bool to mark tombstoned areas), and cross-check the real implementation using random ops.
We can apply upper and lower bounds by chopping up the list of keys/tombstones (we chop at the upper bound on SeekGE, we chop at the lower bound on SeekLT).
Jira issue: PEBBLE-1333
To better test against problems like #5743, we should add a toy testing implementation that deals with small integer keys only (and perhaps use
[]boolto mark tombstoned areas), and cross-check the real implementation using random ops.We can apply upper and lower bounds by chopping up the list of keys/tombstones (we chop at the upper bound on SeekGE, we chop at the lower bound on SeekLT).
Jira issue: PEBBLE-1333