Am 25.11.2016 um 10:58 hat Pavel Butsykin geschrieben: > On 24.11.2016 15:20, Kevin Wolf wrote: > >Visualised, we test these requests: > > > >1: * > >2: ** > >3: * > >4: ******* > >5: ******** > > > >You test inserting the only element, inserting after the last element, > >inserting in the middle and inserting something that overlaps two other > >requests at its start and end. > > > >That's a good start, but it might be worth testing more scenarios: > > > >- Inserting a new first element to a non-empty cache > > What do you mean? To insert an element with zero offset when the cache > already contains other nodes.?
Yes, that would be one way to do it. Maybe just swap requests 1 and 2. > >- Overlapping only at the start > >- Overlapping only at the end > >- Overlapping in the middle (i.e. including existing ranges as a > > subset) > > * With only one node > > * With multiple nodes (like adding offset=2, size=16kb here) > > > > Ok. Kevin
