On Wed, Sep 26, 2012 at 07:17:32PM +0200, Bernd wrote: > 2012/9/26 meh. <meh@xxxxxxxxxxxxxx>: > > > It's not pragmatist at all, it wastes time and resources doing > > replaces when it could have just been really binary and prepend the > > length of the packet, which is the sane way to do something like that > > instead of using an end of packet separator. > > No, these few string replaces do not waste any time or resources, this > claim is totally ridiculous. They do, you have to go over the whole packet to find them and replace them in another buffer or by moving data uselesslessly. The claim that it doesn't is even more ridiculous. > > Also protocol buffers would have been a better choice for something > > like that, it would have beeen even smaller than your protocol and > > easier to implement, it would have made the protocol self-documenting > > too. > > The protocol is self documenting, just print the messages to the > console as they are, the commands mean what their names suggest, it is > so simple you can even simulate a complete handshake and chat session > in telnet manually. Not every hype of the day (xml, protobuf, > binary-json, etc) must be used only because it exists and someone > thinks it has a cool sounding name, sometimes a simple line of text is > enough to transmit a simple foo=bar key-value pair without needing to > wrap it into kilobytes of obfuscating bloat. The name doesn't explain anything about the arguments, the ability to simulate it from telnet doesn't come from the simpleness of the protocol but from the text-based nature of it. Reduncing the amount of redundant data and having a self-documenting and auto-generating definition of the protocol is very useful for this kind of thing. It's less bloated than your "text-based but binary" protocol.
Attachment:
pgpuPVSQvKGVg.pgp
Description: PGP signature
_______________________________________________ tor-talk mailing list tor-talk@xxxxxxxxxxxxxxxxxxxx https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk