- From: <Mike_Spreitzer.PARC@xerox.com>
- Date: Thu, 20 Aug 1998 15:30:03 PDT
- To: ken@bitsko.slc.ut.US
- cc: www-http-ng-comments@w3.org
> > The biggest sticking point is the divergence in type systems between > CORBA and DCOM... > We've been prototyping with BER compressed integers, ... While that's interesting for other reasons, I don't understand why it is a reply to the remark of mine that you quoted. I don't see how the wire encoding techniques you describe addresses the areas where the type systems of CORBA and DCOM have non-trivial differences. More basically, it's not clear to me what you're saying about type systems. I've considered variable-length encodings for integers, but it seems to me that when comparing with fixed-length encodings for fixed-range integer types we have a real apples-and-oranges tradeoff: the variable-length encodings will usually win in bytes-on-the-wire and lose in CPU time (they have to be marshalled and unmarshalled by code with jumps, which cost big on modern CPUs). I do think that dictionary-like representations are useful for composite data in some cases --- I expect the wire encodings of my signature extension ideas will use the dictionary pattern in some places, but not as simply and directly as you suggest. But I don't think it's necessarily appropriate for use in all cases.
Received on Thursday, 20 August 1998 18:30:17 UTC