> I'm thinking that we should always send the "cursor"(also from API),
> having a "zero" cursor would then simply mean "start scan".
>
> Also looks like DbinfoScanReq and DbinfoScanReq should have exactly the
> _same_ definition -> we can probably remove the "struct DbinfoScanCursor"
The idea is that the cursor should be opaque for the ndbapi,
that the ndbapi should simply take what's in there and copy it into the "next-req"
wo/ knowing what's in there...
And if the cursor is opaque from ndbapi's perspective...it can't create a "zero" cursor.
---
The cursor is being opaque, so for future upgrade-ability.
Maybe we should also add a "cursor-sz", so that ndbapi could handle the cursor having a
different size...
---
Summary: i'm not super keen on "zero" cursor idea
/Jonas
