On Wed, Oct 10, 2001 at 02:22:51PM +0200, Oskar Sandberg wrote: > On Wed, Oct 10, 2001 at 11:07:23AM +0100, toad wrote: > > But what clients will do, most likely, is request all pieces and use whatever > > comes first, improving latency. > > Which for all intents and purposes is fine. The real problem is if all > clients only requested the first parts by default.
Of course, if a code is used, then the nodes *must* request parts randomly until they have enough pieces to form the entire file. Otherwise, as you point out, Freenet's automatic mirroring and deletion will work to make the first k parts very availible and the last n-k parts will be hard to find. This will destroy the assumption of your code. PS: In this discussion what about the MetaData? Doesn't one always need to download the metadata precisely? PPS: couldn't this all have been done in "userspace"? One could easily make a tool that would allow you to do splitfiles in 0.3 by downloading some MetaData file which describes where to get all the pieces. Oscar. -- [EMAIL PROTECTED] http://pobox.com/~boykin ICQ: 5118680 Key fingerprint = 159A FA02 DF12 E72F B68F 5B2D C368 3BCA 36D7 CF28
msg02885/pgp00000.pgp
Description: PGP signature
