Actually, I think that the current method is so clumsy, that if people are not going to accept the typing of fields I will just throw the entire extra layer out, and do the encoding and decoding of values in the code that reads the fields when the messages are handled (ie where it doesn't belong).
Since all presentations (to follow Theo's names) would then have to carry all data as strings, so it wouldn't matter. It's your choice people, personally I find it a shame to screw something so nice over people's refusal to add one character to the serialization, but that is compromise I guess. On Sun, 14 May 2000, Brandon wrote: > > Nodes pass fields they do not understand as a matter of forward > > compatibility - > > it makes it possible to add fields that every node does not have to obey > > without worrying about every single node implementing them. > > But if we only drop unknown fields are a protocol gateway then we'll only > lose forward compatibility at a protocol gateway, and only when the > protocol gateway is old. If the protocol gateway is up to date, then we'll > still have forward compatibility on the other side. I don't think that the > number of out of date protocol gateways is going to be great enough that > it will be problematic. > > > > _______________________________________________ > Freenet-dev mailing list > Freenet-dev at lists.sourceforge.net > http://lists.sourceforge.net/mailman/listinfo/freenet-dev -- ___ Oskar Sandberg md98-osa at nada.kth.se _______________________________________________ Freenet-dev mailing list Freenet-dev at lists.sourceforge.net http://lists.sourceforge.net/mailman/listinfo/freenet-dev
