If we redid Netrek from scratch today, I’m confident we’d use a type-length-value encoding. That way clients and servers can parse past new types of messages without corrupting the data stream. Darrell Sent from my iPad > On Mar 26, 2021, at 2:05 PM, James Cameron <quozl at laptop.org> wrote: > > On Fri, Mar 26, 2021 at 01:53:52PM -0500, Alec Habig wrote: >> James Cameron writes: >>> >>> Tedd Hadley and Heiko Wengler's amazing short packets code from 24th >>> May 1993. ;-) >> >> Original motivation: Heiko was trying to play from the end of a >> network connected by string, tin cans, and carrier pigeons from >> rural Austria. > > Indeed. In my case, I was playing over 2400 baud modem running SLIP > into a VAXstation. Short packets made all the difference. > >> short packets made enough of a difference that we could field a (semi) >> competitive team from Europe in the INL for most of the seasons the INL >> was a thing: the Eurotwinks. >> >> Our home games were great, because the NA based players had no idea how >> to play with 300ms ping, but that was the way we played all the time. >> Even for home games for most of the team :) > > Yes, I remember how players with low latency showed a different > playing style and were still vulnerable to the attacks from players > with high latency. > >> In related news, the network stack I stole from Heiko (and netrek) to >> run the SNEWS supernova early warning system has also been showing its >> age, for the same packet-length confusion reasons as started this >> thread. We're redoing it all in python and kafka now. > > I'll contact you offlist about that. > > -- > James Cameron > https://quozl.linux.org.au/ > _______________________________________________ > netrek-dev mailing list > netrek-dev at us.netrek.org > http://mailman.us.netrek.org/mailman/listinfo/netrek-dev