Re: HTTP version 1.0 or 1.1?

Roy Fielding:
[....]

>Since the IETF recently created the BCP ("Best Current Practice")
>series, we can actually do this now within a reasonable framework.
>
>The action required would be for Henrik and I to create a stripped-down
>HTTP/1.0 document and release it along with the current draft renamed
>as HTTP/1.1.  This will take three days for us to accomplish, but may
>be worth the saving in confusion and aggravation.
>
>Does the WG want to do this?

I see a number of advantages to this approach, and almost no
disadvantages, except that it may take rather more than 2 times 3
person-days, if the BCP document is really meant to reflect the amount
of http functionality that can be relied upon to be present today.

In one area I am familiar with, redirection codes, best current
practice would be `only 301 and 302 can be expected to work as
specified, and only in responses to GET requests' (at least, that is
my observation on doing experiments with browsers).

Current support for redirection is rather less than what is specified
in the 1.0-01 draft, so this points to the need for a BCP document.
On the other hand, the redirection case shows that, to write a good
BCP, a lot of information about the behavior of current web software
will be needed, and this information may take a long time to get
(either by doing experiments yourselves or by asking for information
on www-talk).

And of course, as soon as the BCP is released for review, you will get
flamewars (i.e. large amounts of wasted time) on www-talk and http-wg
along the lines of `well, I know that browsers X, Y, and Z do not
support protocol feature P, but I call X, Y, and Z broken browsers
because of this, so P needs to be in the BCP document nevertheless'.

> ....Roy T. Fielding  Department of ICS, University of California, Irvine USA

Koen.

Received on Saturday, 12 August 1995 04:01:22 UTC