On Tue, 1 May 2012, Jeff King wrote: > On Tue, May 01, 2012 at 01:17:03PM -0400, Nicolas Pitre wrote: > > > However, this is possibly a good baseline to determine a default value > > for window-memory though. Given your number, we clearly see that good > > packing can be achieved with relatively little memory and therefore it > > might be a good idea not to leave this parameter unbounded by default in > > order to catch potential pathological cases. Maybe 64M would be a good > > default value? Having a repack process eating up more than 16GB of RAM > > because its RAM usage is unbounded is certainly not nice. > > Would that preclude a 65M object from being delta'd at all? If you are > putting a limit of 64M and it means we look at 50 delta candidates > instead of 60, then that is probably not going to hurt our size too > much. But if you have large objects that do happen to delta, the > difference between looking at 0 and 1 objects could have a big impact. If I remember correctly (long time ago since I wrote that code) there is always a minimum of one object in the search window. So with 32M objects you'd have only 2 candidates, with 33M objects and bigger there would be only one candidate. Obviously the window could still be filled with smaller objects if the total object size is less than 64M. Nicolas -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html