On Sun, Jan 26, 2003 at 04:03:01PM -0500, Gianni Johansson wrote: > On Sunday 26 January 2003 13:51, you wrote: > > > > > > Well, I thought you were worried about something running in Fred's JVM > > > > by default. > > > > > > When you next talk to the press and mention Freenet as a replacement for > > > internet radio (which it is not), everyone will turn it on and try to run > > > it for a while until the network collapses. > > > > More groundless pessimism. You have no evidence that Freenet could not > > deliver an FM quality stream - in fact, we have seen indications that it > > can, and things are only likely to get better as the network improves. > > I am not saying that it will never be possible, it's just not a reasonable > thing to encourage people to do now given the current state of the network. > If fish wants to work on this that's great. > > Freenet is better than internet radio. You should be able to write a client > that prefetches and concatenates files from a dynamic playlist and presents > them over HTTP. That would play to Freenet's strength's much more than > encouraging people to re-insert the same data over and over via "streaming > insert clients". > > > > > > You don't want fred to have to compete with code that is running > > > continuous FEC decoding in the same JVM. Perhaps your machine is so > > > powerful that you haven't noticed, but we can't get fred to run with > > > reasonable CPU usage *all by itself*. > If you really care so passionately about "streaming", what about doing some > work to figure out what's going on with CPU usage? Hey, I tried to do some profiling. Java was a bit of a pig about it, only letting me do short profiles, but the tentative results suggest that crypto dominates. > > [point for point rehashing of take no hostages comp sci debate on stability > and encapsulation verses efficiency omitted. ] > > I don't appreciate your tone and ad hominum attacks. > I have limited time to devote to this project and the amount that I would > waste engaging in this kind of debate is directly subtracted from it. So I > refuse. Declare utter total and devestating victory if you will. I have code > to write... > > > > 2)If all requests go over FCP fred can better defend itself from > > >clients that make unreasonable requests. > > Again, this is a meaningless statement. We can't actively prevent > > people from running one client or another with their own node (hey, > > maybe with Palladium we could...?!). > All I meant is that when the FCP server is overwhelmed, it can stop answering > requests. > > --gj > > p.s. > Fish, I am not out to get you. Testing in fred's JVM is fine. As far as > widespread deployment goes I guess we will cross that bridge when we get to > it. This all seems quite meaningless without some hard data.
-- Matthew Toseland [EMAIL PROTECTED][EMAIL PROTECTED] Full time freenet hacker. http://freenetproject.org/ Freenet Distribution Node (temporary) at http://amphibian.dyndns.org:8889/RMMHBxLOISw/ ICTHUS.
msg06410/pgp00000.pgp
Description: PGP signature
