>>>>> "MS" == Melvin Smith <[EMAIL PROTECTED]> writes:
MS> That answers my question of overhead with regards to XML headers. MS> If there is a single header for defining the type of "stream" then the MS> actual serialization can be dense enough.
MS> I just needed clarification. :)
well, that was my take on what dan said. i could have misinterpreted him.
MS> I think a grammar should be developed for this, since it should likely MS> be implemented as a recursive parser.
there are many ways to encode serialized stuff and not all are conducive to grammars IMO. some may be simple type/length/binary value things which are best expanded in nice fast loops. other formats could just go into pure ugly XML which can be parsed back into data. also dan mentioned the half-frozen human readable types (which both the binary AND XML formats need!) which don't (or may not) get parsed again.
When I say grammar, I don't mean as input to a parser generator. I'm accustomed to using grammars for any sort of non-trivial packet or data format. It makes dealing with and discussing the format much easier.
-Melvin
