Closed fsteeg closed 9 years ago
size
has no impact when using scroll
. So there is no need for validation.
executeScrollScan
calls bulk
, which calls startInitialResponse
, which calls search
, which sets the size on a request and executes it:
Ah, yes. The first lookup uses indeed size
. It is done to get the overall hits. Best would be to hardwire the size to 1
when doing scroll
. Doesn't matter much - merging an closing.
The new chunked response uses a different code path that skipped parameter validation. This allowed requests with unlimited size, see included test for an example.