- From: Daniel DuBois <ddubois@rafiki.spyglass.com>
- Date: Thu, 7 Dec 1995 13:18:18 -0600
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
I have a few concerns I'd like to address about the current state of the HTTP working group. I tried posting these Tuesday, but by mailing software is malfunctioning. First I'd like to plead for someone to give an overview of what happened at the night meeting of the HTTP WG. Unfortunately, I had to leave early, and the idea of waiting two months for minutes is unpleasant. As such, I have some feelings on the last minute agenda change at the Dallas IETF meeting. 1) I think the addition was ill-timed. The entire Spyglass group had plane reservations and hotel arrangements that required we leave in the late afternoon. This is a small gripe, but we at Spyglass, myself in particular, really, really wanted to participate, mainly because of the issues I'm going to address later. 2) The structure of the Agenda that pushed the bulk of the HTTP 1.1 to the 2nd session was horrible! No way should we have spent the 1st session discussing the tunneling/session-ext/payment/etc drafts. The HTTP 1.1 spec is THE issue of this group. The next thing basically is a process issue. Some of the informal hallway conversations I had confirmed my suspicions that the current progress of the group is insufficient. As I see it there are three things working together to impede our progress. If only one of these things existed, it would be perfectly acceptable, but together they spell doom: 1) the addition by the editor of large surprises to the HTTP 1.1 draft without any discussion in the working group, 2) the very infrequent revisions of the spec, 3) the monolithic-kitchen sink nature of the document. Now before you say: "Putting things in the I-D is the way to submit an idea for consideration by the group!", I must reiterate: that works fine if you have frequent revisions on a small doc and no one is actually using the draft spec. People are trying to code to these documents. The HTTP WG is not responding to the needs of the marketplace. The addition of Logic Bags, PATCH methods, two-stage responses and the like without both the input of the group, and constant revisions of our "main product" is just not workable. I am calling for a rough consensus from this working group that the best way to progress on HTTP 1.1 is to a) require any large additions to the spec be discussed in this group - as a separate Internet Draft if convenient - before any introduction, and b) we commit ourselves to a somewhat only-what's-necessary Draft of 1.1 to get some kind of consensus in a reasonable amount of time. There was also a sentiment in Dallas that not only should big chunks not be added, but that the HTTP spec should be broken up, and the more I think about it, the more I have to concur. I think I'll separate individual gripes about particular aspects of the spec to a separate message, and I'll propose some potential break-outs there. ----- Dan DuBois, Software Animal http://www.spyglass.com/~ddubois/ I absolutely do not speak for Spyglass.
Received on Thursday, 7 December 1995 11:28:45 UTC