Stef's suggestion
klensin@infoods.mit.edu Tue, 06 October 1992 14:58 UTC
Received: from ietf.nri.reston.va.us by IETF.NRI.Reston.VA.US id aa03044; 6 Oct 92 10:58 EDT
Received: from NRI.RESTON.VA.US by IETF.NRI.Reston.VA.US id aa03038; 6 Oct 92 10:58 EDT
Received: from dimacs.rutgers.edu by NRI.Reston.VA.US id aa08980; 6 Oct 92 10:58 EDT
Received: by dimacs.rutgers.edu (5.59/SMI4.0/RU1.4/3.08) id AA14244; Tue, 6 Oct 92 10:35:30 EDT
Received: from fred.plymouth.edu by dimacs.rutgers.edu (5.59/SMI4.0/RU1.4/3.08) id AA14240; Tue, 6 Oct 92 10:35:27 EDT
Received: from PSC.PLYMOUTH.EDU by PSC.PLYMOUTH.EDU (PMDF #12261) id <01GPMIYV39V48Y5VTN@PSC.PLYMOUTH.EDU>; Tue, 6 Oct 1992 10:38 EST
Date: Tue, 06 Oct 1992 10:38:00 -0500
Sender: ietf-archive-request@IETF.NRI.Reston.VA.US
From: klensin@infoods.mit.edu
Subject: Stef's suggestion
X-Orig-Sender: JKLENSIN@psc.plymouth.edu
To: ietf-smtp@dimacs.rutgers.edu
Message-Id: <01GPMIYV39V48Y5VTN@PSC.PLYMOUTH.EDU>
>Take the new draft documents (EHLO, 8BIT, and SIZE) as concrete >proposals, and accept or reject them. Stef, An interesting suggestion, but one that I'm quite hesitant about. The difficulty I see, as I think we discussed privately some weeks ago, is that things get lost in this process. I have no problems with their being dropped, but I think the WG needs to do that explicitly, rather than just losing them. Since I haven't been able to study Keith's latest draft yet, and suspect that the WG has reached consensus on the general extensions model in draft-rose-extensions-03.txt, let me draw some examples from the 8-bit area. (1) This proposal changes the "8 bit transmission" model from one of "client asserts that it is about to send a particular message containing 8 bit chars and gets server verification" to "server asserts that it can handle 8bit messages from or two any source". A great deal of discussion during the WG's development period seemed to point to the necessity of the former, especially for gateways into other environments, and that issue has been raised again in the light of more recent discussions. I don't think it is reasonable to lose that position; the WG needs to explicitly take a stance on it. (2) This proposal changes the moderately detailed specification for trace fields in RFC-ZZZZ to an overloading of the 821/822 "Received:...with" phrase. But it doesn't [re]define what is to be used with "with" in saying "e.g., ... with binary-to-base64" (middle of page 5), nor does it deal with the issue of registering these values (required in the existing RFCs, although no registrations had occurred as of the last time I asked Joyce). It also changes the semantics of "with" from notions about protocol or transport mechanism to conversion assertions. Again, I don't have any real problem with this, but do know, from long experience with gateways that have to perform conversion, that being specific about what is expected here (and expecting a lot) are quite important to any sort of debugging. So, if the WG wants to let that additional documentation and specificity go, then it should explicitly make that decision. (3) The second paragraph under 4.1 appears to explicitly legitimize the sending of untagged (non-MIME) 8bit messages. On the other hand, it appears to prohibit the use of externally-available information (e.g., from agreements among consenting adults) to properly tag the converted object. We know that most untagged 8-bit stuff will be text of some sort; we also know that there will be many cases in which the character set in use is perfectly well known, e.g., at what RFC-ZZZZ and previous WG discussions describe as an "enclave boundary". RFC-ZZZZ prohibits untagged 8bit messages, after long discussion about the interoperability threats of letting them (encouraging them to) float around. It also makes explicit provisions for accurately tagging them should they occur, again after long WG discussion. Again, if the WG wants to abandon those discussions and decisions and do that, it is fine, but the decision should be explicit. My guess, Stef, is that, for better or for worse, we are going to have to reach some explicit agreements and then either modify the new draft documents and write a new one or two or we are going to have to make another pass through RFC-ZZZZ to incorporate some or all of the protocol changes implied by the new documents. Even if the WG goes with "prefer to follow Stef's alternative", I know I've got enough editorial quibbles with the new drafts to require at least one more iteration and I assume that others do also. Either way, while I don't see "a long time" in the future of getting a WG product out, I don't see "this week" either. ---john
- Stef's suggestion klensin
- Re: Stef's suggestion Einar Stefferud