RFC 2616, partitioned

I spent much of the last month fiddling with RFC 2616 and the
various ways that it could be partitioned into the original design
layers of the protocol.  The result is at

   http://labs.apache.org/webarch/http/draft-fielding-http/

and demonstrates that the protocol can be described in smaller
documents without requiring cross-references all over the place.
In fact, the only cross-references left are editorial in nature
or places where caching tries to redundantly define other
aspects of the protocol.

However, it is far from being readable at this stage.  In fact,
it shows how little of the operational aspects of the protocol
were described in 2616 due to the pressure not to make the big
document any bigger, and how frequently the cross-references
in 2616 point to sections that have little (if any) relevance
to the discussion.

I was trying to minimize changes from RFC 2616, so haven't
rearranged the sections where they would logically belong (aside
from one example of chunked decoding pseudocode that was moved
from an appendix to the one section that referred to it).
Before doing any further structural changes, I want to commit all
of the editorial changes from the errata (as tracked under the
link to DF_Issues).  I will be too busy over the next 2.5 weeks
due to travel and an office move, so now would be a good time
to poke holes in the concept, criticize the structural breakdown,
or point out additional errata that apply to both this and the
rfc2616bis document that Julian is editing.

BTW, Julian also has write access to Apache Labs stuff, though
he may not know it yet.  If anything is changed, please be sure
to follow the previous examples in commit logs and jira issues.

....Roy

Received on Wednesday, 17 October 2007 02:03:56 UTC