Talk:SE250:lab-5:hpan027

Comments from John Hamer

• Nice justification of your choice of sample size.

• Perceptive comment on the low/typical spread.

• 1000/10000 — this is a very lightly loaded hash table. An explanation of your method and reasoning would be in order.

• There are lots of curious things going on here. You should point to specific instances. For example,

• Java_Object_hash low 1000/10000: llps = 1000

every string hashed to the same bin! Yet it does well on the stats tests

• Java_String_hash low 1000000/100000: llps = 18, expecting 26.6

Now it's doing better than expected

• Why no Java Integer hash?

Overall, you show very good understanding of the material and have presented your results intelligently.