[mnet-devel] BW unit tests

Zooko zooko at zooko.com
Sat Feb 15 15:48:21 GMT 2003


 myers wrote:
>
> Why not move the handicapper stuff into Node?  Or are you saying that the
> handicapper is still needed by egtp for network ops, and that we should
> invent something new for the application layer?

I'm saying that the abstraction of "handicapper" should go away as fast 
as possible and be replaced by customized decision-making for each 
specific job.

EGTP has no need of decision-making at all, except for choosing your 
relay server, and the current RelayListener already does that in a 
customized way.  (It might still use handicapper, but if it does it 
doesn't really need it and we can remove it.)

BlockWranglingStrategy.GSR and .KISS currently use handicappers, but 
would probably be faster and smarter if they didn't.

Search currently uses handicappers, but the new search that I'm about to 
commit doesn't.

So AFAICT as soon as someone writes a clever blockwrangling strategy 
that doesn't use handicappers, then we can extirpate the remaining 
handicapper code from _new.

--Z




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
mnet-devel mailing list
mnet-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mnet-devel




More information about the Mnet-devel mailing list