Re: safe PUT

" at Dec 29, 95 01: 42:15 pm
Date: Mon, 08 Jan 1996 00:03:11 -0800
From: Dave Long <dave@geek.navisoft.com>


Larry asks:
>We've had a long discussion with various counter-proposals floated,
>but not much convergence. I'll ask all of you to decide between one of
>the following three alternatives (I can think of no others):

>- Should PUT be removed from HTTP/1.1?

No - The original specification is better than nothing.

>- Should PUT go into HTTP/1.1 as originally specified,
> but with a warning as to its unreliability?

Yes - I am happy with the original specification; it does
have weak points, but it is not hopelessly broken.

>- Are you interested in drafting a counter-proposal?

Maybe - I would like to see the following added:

	Clients may avoid sending the remainder of a request
	at any time after the response has started.

However, before I attempt to make a formal counter-proposal,
I'd like to have some idea that others (Roy? Roger?) have tried
this solution and it solves their problems as well as it does mine.
I can make Windows, Sun4, Solaris, HP, and Mac clients available
for testing.

Summary
-------
If there is enough interest in HTTP-WG land, I would like to
avoid sending entity-bodies unnecessarily.  If there isn't,
then I'd much rather have PUT in HTTP/1.1 as originally specified
than to have it left out.

-Dave

Received on Monday, 8 January 1996 00:07:33 UTC