Block size optimization - let rsync find the optimal blocksize by itself.

[Olivier Lachambre]

At 12:36 03/07/2002 +1000, you wrote:
>On Tue, Jul 02, 2002 at 11:06:30AM +0200, Olivier Lachambre wrote:
>> At 09:19 01/07/2002 +1000, you wrote:
>> 
>> >[...]
>> >This relies on optimal block size being related for a set of files. I'm not
>> >sure that this heuristic actually applies, and I don't know how much benefit
>> >this would buy for all the complexity it would add.
>> >
>> 
>> I think that many clients do not care about multiplying complexity by 2 or
>> 5, even if the speedup rate is only multiplied by 1.1.
>
>The only problem is, multiplying complexity by 2 or 5 multiplies the bugs by
>2 or 5. Are those clients still excited by the 1.1x speedup now?
>
>The other problem with increasing complexity is the "setting" effect that
>complexity causes. The more complex code becomes, the harder it gets to
>modify, so that 1.1x speedup that introduced 5x complexity now makes it
>impossible to implement a much better 1.5x speedup.
>
>In my experience, complex solutions usually have worse performance than
>simple ones anyway.

When talking about complexity I mean complexity as defined in theorical
computing (that's at least how we call it in French). In this case,
complexity is realated to the time the algorithm needs to be executed, 
and not to the size of the source code or the size of the binary.
For example :
     for i = 0 to n do
       for i = 0 to n do 
         (nothing)
       done
     done
has a complexity proportionnal to n^2, which is sometimes a lot, but the
program is not complex in any way.
Myself at home I would be happy to spare 10% of the time spent on the
Internet (as local communication are not free) but I don't care if my 
computer works twice or 5 times more, since I have a 56K modem. I guess that
the problem for many user is low bandwidth and not low CPU ressources
(I know there are exceptions). I think that the code of the algorithm 
I proposed would not be simple but would take time to run.

Olivier
__________

Olivier Lachambre
2, rue Roger Courtois
25 200 MONTBELIARD (F)

<lachambre at club-internet.fr>