- From: Claudio Allocchio <Claudio.Allocchio@garr.it>
- Date: Tue, 20 Nov 2001 18:21:40 +0100
- To: discuss@apps.ietf.org
- cc: Brian E Carpenter <brian@hursley.ibm.com>, john_ibbotson@uk.ibm.com
First of all, the draft seems clearly about client/server communications,
and thus using the term "messages" might be the origin of a dangerous
confusion. Instead of "messages" it would be better to talk about
"client/server information exchange". A "message" is very ofter related
to "global messaging services", which is defintly something different.
Then, IMHO, the idea in general seems to me "yet another misuse of HTTP".
I tend to be worried when I see the approach of using HTTP to do
anything, and WEB browsers as a good for all application interfaces. HTTP
il Hyper Text Transfer Protocol (with a number of syncronization and
reiability problems we all know), and it is not the general all purpouse
transport protocol. It is perfectly true, Keith and Mark: it is an
application protocol.
Tha overall analysis of the requirements is fine, but conveying the idea
that HTTP is the solution make me strongly disagree. Internet
applications should NOT be tied to HTTP, and should eventually use their
own transport / application protocol, carefully desinged to fulfil the
application needs. Playing any possible method to force an application
and its protocol into HTTP and a WEB browser is something we should
simply avoid as much as possible....
just my two cents...
------------------------------------------------------------------------------
Claudio Allocchio G A R R Claudio.Allocchio@garr.it
Project Technical Officer
tel: +39 040 3758523 Italian Academic and G=Claudio; S=Allocchio;
fax: +39 040 3758565 Research Network P=garr; A=garr; C=it;
Received on Tuesday, 20 November 2001 12:22:35 UTC