- From: Chris Newman <chris@innosoft.com>
- Date: Thu, 11 Feb 1999 15:57:09 -0800 (PST)
- To: Jim Gettys <jg@pa.dec.com>
- Cc: discuss@apps.ietf.org
On Wed, 10 Feb 1999, Jim Gettys wrote: > I don't want to be in a situation where HTTPng is believed to be the "one > size that fits all", even if HTTPng works out (and its not soup yet), > and would want it free to optimize to web operations, which have a different > set of constraints than many other applications protocols: e.g. low latency, > since a user is driving it is important for interactive feel, which is > NOT a requirement on, say, mail. It is those kinds of requirements (along > with the "brittleness" of current systems like CORBA), that have made > me believe something other than CORBA or Java RMI is needed for the web. > > It is also not clear to me that HTTPng requirements should bind APPLCORE > hands, as the requirements may be found to be disjoint. > > While the web is a common application protocol, it should not, > in my opinion, be the only one. I agree with all of this. Perhaps APPLCORE and HTTPng will have sufficiently similar components that they should be aligned for code re-use purposes. Perhaps APPLCORE should target simple low-end protocols akin to POP/SMTP/etc and HTTPng target more complex high-end protocols (or even just web browsing) and they'll be separate but useful tools for our toolbox. - Chris
Received on Thursday, 11 February 1999 19:03:39 UTC