On Saturday 03 May 2008 15:01:30 Alexander Lehmann wrote: > I think the rationale behind polling ahead on the hash slots was that the user may miss an inserted key sometimes, if this is the case, it would be useful to insert the message to a new slot when starting a new rts, it that cannot happen, it is not necessary since the old key can be used. > > However if we do not expect this to happen, do we really need to poll ahead on the keys?
Now you come to mention it, it ought not to be necessary - pehaps we can remove it, or at least decrease it. As far as I can see, the only reasons we'd find messages on a slot having not found any on a previous slot would be if there were a problem with sngle-recipient keys on Freenet (in which case we're screwed anyway) or if there's a bug. I think some backwards incompatable changes to RTS messages will be in order in the near future, so I think we'll be gaining a 'development version'. Perhaps we could disable poll-ahead in the dev version? Dave > > > Message: 2 > > Date: Thu, 1 May 2008 16:23:59 +0100 > > From: Dave Baker <dbkr at freenetproject.org> > > Subject: Re: [Freemail] Log: Don't re-insert mail to different slots > > To: freemail at freenetproject.org > > Message-ID: <200805011623.59719.dbkr at freenetproject.org> > > Content-Type: text/plain; charset="iso-8859-1" > > > > On Thursday 01 May 2008 16:01:47 Alexander Lehmann wrote: > > > I think it may be useful to first try to insert the message to a new slot > > > (in case the user has missed a slot somehow) and later to retry rts or is > > > this a situation that doesn't seem likely. > > > > > > freemail-request at freenetproject.org wrote: > > > > Modified: > > > > trunk/apps/Freemail/src/freemail/OutboundContact.java > > > > Log: > > > > Don't re-insert mail to different slots - re-send the RTS instead. This > > > > will allow contacts to be re-established if the inbound contact has been > > > > deleted. > > > > Well, I was thinking along the lines that the user missing a slot shouldn't > > happen, since we'll only advance through a slot if we find a message on it. I > > don't think it's every likley to be that a user cann't fetch one ket but can > > fetch the same data on a different key. > > > > Feedback welcome though. > > > > > > Dave > > > -- > Alexander Lehmann <alexlehm at myrealbox.com> > > > _______________________________________________ > Freemail mailing list > Freemail at freenetproject.org > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/freemail >
