Re: Major errors in Caching and Cache-Control

>
>
>Date: Wed, 05 Jun 1996 15:36:19 -0500
>To: "David W. Morris" <dwm@shell.portal.com>,
>        http working group <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
>From: Daniel DuBois <dan@spyglass.com>
>Subject: Re: Major errors in Caching and Cache-Control 
>
>>I agree, no-transform should exist (and I see no reason to forbid 
>>transformation with HTTP).
>
>You mean other than the fact that it blows semantic transparency out the window?

Semantic transparency is an ideal, to be relaxed for many good reasons.
I believe the current text even talks about it in those terms.
Doesn't bother me at all.  The reasons seem good.

>
>How about we go ahead and leave no-transform in, but include a Note: below
>it that includes a statement to the effect: "Although HTTP does not
>currently allow intermediary proxies to transform responses on their way to
>the requestor, future versions of HTTP may allow such behavior under certain
>circumstances, and the no-transform directive is included now in
>anticipation of such behavior."
>
>At least, that was what I was going to say until received Benjamin Franz's
>message while typing this.  The more I think about it, the more I agree with
>the sentiment below:
>
>>If this allowed at all, it should be in the affirmative mode: 
>>transform-allowed. The default without question should be no content
>>transformations allowed unless explicitly stated otherwise. At no point
>>should intermediates be allowed to transform content unless I
>>affirmatively give them permission to do so.
>

The affirmative doesn't do any good.  Current practice (that I was not
aware of until this thread took on a life of its own; I guess I should
have not been so naive) is that transformations are occuring.  So
saying that you can transform something is a no-op; current practice
is that you can transform it, so saying so again doesn't help.

Since the horses are galloping a long way from the barn already, at
most we can do is clean up the mess and forbid it in the future.
Seems like all we can do (independent of what would be nice) is
express the negative; you can't transform specific data, and then the
horses get back in the barn someday, when HTTP/1.1 is widespread.

				- Jim

Received on Wednesday, 5 June 1996 14:11:51 UTC