On Thu, Dec 02, 2004 at 07:54:45AM +0100, Someone wrote: > ~From your posts I guess you haven't been running a stable node for some > time, > then installed a new one and tried to insert your flog the same day. Then it > doesn't suprise me that it took quite some time. Let your node run a week > with > frost (to train it) and you will see a great improvement, shure a week is a > long > time, but if you take in mind that it has to learn a whole network and find > some > good routes/nodes it's not that much. Since build 5099 I see specialization > on > my node and insert speed is very good. I currently run 4 freesites, the > smallest > about 100kB, the biggest nearly 30MB. The smallest is usually inserted > within > 10 Minutes. The biggest takes between 4 to 6 hours, this isn't too long for > the > amount of bandwith my node is allowed to use.
This is IMHO a major problem for the project at present: the learning time. I don't think that this will be really solved short of fixed size small chunks. But we have some way to go before we can do that; I don't think we should do it before 0.6. > > | And as for never bringing any alternatives...yeah, right. I remember > clearly > | I made a whole list of suggestions and alternatives for improving the > | end-user experience. Seems that memory is very selective. Why, only a few > | days ago, I pointed to an error in the uninstaller. But hey, feel free to > | ignore all that, and say I whine (not without reason in any case) about > the > | fact that freenet still sucks for the most part, mostly because of bad > | directions that have been taken. Didn't I suggest to go for UDP and small > | chunks, and leave the NGR (that, after all this time, still hasn't > actually > | proved it worked) if the simulations don't clearly show a benefit? Yes, I > | did. But that, I suppose, is rapidly forgotten, even it's exactly the > | alternatives that you claim I never gave. It's easier to say I 'just > whine', > | so you can regard me as a troll, instead of contemplating if I don't have > | valid points. > > Pointing out errors and making valid suggestions is a different thing. As > for > going for UDP, this actually would have made it worse. I know you suggested > it to push holes through NAT routers, but this doesn't work on most routers > here. And this for a very simple reason, almost all (yes even the low end > cheap trash ones) routers you can get here have full SPI. So it isn't > possible > to push a hole into their NAT, what's even worse is that the cheaper ones > have SPI, but only allow to open or redirect TCP ports, not UPD ports. So > going > to UDP would have ruled out users with one of these routers, even if they > knew > how to open/forward the listen port. The only thing that might help with > this > is to support UPNP. But it wouldn't change much, because ppl always get told > that UPNP is a security risk, and they tend to believe it and disable it. There are a lot of different ways to support hole-punching through firewalls. Simultaneous connect is one option, UP&P is another, UDP is a third. Some of these will eventually be implemented, but are not a priority at the moment. > > As for small chunks, IIRC it was suggested by Toad not by you. You only > agreed > that it might be usefull, just like I did. > > And I also did run a node pre NGR, and I saw a great improvement when > switching > it over to NGR last year. Really? Interesting... > > | In neither case it validates deleting my account, so that I now would have > | to change all references and links to it into new ones, if people want to > | contact me via the site or the wiki I run. Oh, yes, but that's all not a > | 'valid contribution', that's right. > > Oh, how bad. You have to edit some lines of HTML or change some entrys in a > DB, > this will be the end of the world. :-P -- Matthew J Toseland - [EMAIL PROTECTED] Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so.
signature.asc
Description: Digital signature
_______________________________________________ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
