Stateless IM Continued

[Alex White]

Andy Pieters wrote:
> Hi list
> 
> Thank you all for your suggestions.
> 
> I have researched your answers and I thank you for your suggestions.
> 
> The main thing would be that if there is a glitch in the connection between 
> the two computers that the "sender" just tries to send again after an 
> interval of a few seconds.
> 
> Yesterday I was talking to my dad via icq (he's using windows icq client) and 
> I use Kopete.
> 
> There was a thunderstorm going on and there were quite a few connection breaks 
> of a few seconds each.
> 
> The connection with the icq server got severed each time I saw lightning but 
> the most anoying part was that I didn't get any warning of connection 
> failure.
> 
> Basically I had to connect, type a sentence, see if an answer gets trough, and 
> if not, disconnect, and connect again to see the answer.
> 
> If it were peer to peer with each side retrying to send we could have 
> continued our chat without even noticing the glitches.
> 
> We are about 1.500 km (~1000 miles) apart from each other.
> 
> 
> With kind regards
> 
> 
> Andy
> 

I'm bad at trimming so won't, as the entire thing is relevant. I
only think that tcp is going to help you there, because udp doesn't
care if the packet got there or not.

However, I don't know of any technology aside from starting up a
gnutella client of some sort? IRC would probably be the way to go
though, as that would immediately disco if there was any severing of
a connection. Most clients (xchat I know for certain) will reco all
on its lonesome if it loses connectivity. And as was mentioned
before with direct connect via DCC you have at least that much
security but I don't think that's really secure since if either of
you are being sniffed all that data is (I think) plain text.

HTH

Alex White