On 05/08/2008 11:31 AM, Greg Hudson wrote: > On Thu, 2008-05-08 at 11:40 -0400, Stephen Pendleton wrote: >> I think the reasons are: >> >> - There is no reason have have stream restarts because the reasons >> they were introduced ended up not to be correct. >> - The protocol "looks" better. Fewer stanzas are needed during the >> login/authentication process I believe. >> - Some implementations will not need to throw away/reset their parsers >> anymore. Currently this is necessary for all implementations that I >> know of and those using the style of architecture I show below will no >> longer need to do this because the XML parser will be presented with a >> complete XML stream, instead of multiple ones. I don't think we should >> be changing the core protocol based on implementations though, so this >> is less important to me. > > These are not compelling reasons to introduce a change to a protocol > already in wide use. You cannot "clean up" an existing protocol, only > make it messier by introducing transition issues.
I have not read this entire thread yet, but I'd like to point out that the possible change is not a modification to the existing behavior, but a new yet slightly modified stream feature. We would not change the behavior of STARTTLS and SASL negotiation as currently defined. Peter -- Peter Saint-Andre https://stpeter.im/
smime.p7s
Description: S/MIME Cryptographic Signature
