Re: Strange definition of Frame in Cougar DTD

David Perrell (
Wed, 30 Apr 1997 09:56:18 -0700

Message-Id: <>
From: "David Perrell" <>
To: "Arnoud \"Galactus\" Engelfriet" <>,
Subject: Re: Strange definition of Frame in Cougar DTD
Date: Wed, 30 Apr 1997 09:56:18 -0700

Arnoud "Galactus" Engelfriet wrote:
> It is also possible that user agents allow the user to _disable_
> frames support (Opera 2.1, MS IE for Mac, for example). Then the
> browser would effectively be "downgraded" and needed to display

If the rule were to display NOFRAMES content when no frames have been
set, this would not be a problem.

> Which frames-capable browsers *do* handle NOFRAMES as the working
> draft suggests, anyway?

MSIE 3.02 Win: yes. NSN 3.01 Win: no.

> Ok, how do I detect that on the server? Don't you mean client-side
> scripting?

Strictly speaking, you can't. All late-model UAs are identified in an
HTTP header variable. So content can be delivered according to features
supported by the UA. You point out the problem with this method:
supported features can be disabled.

Would it not be better to have an HTTP variable "HTTP_FEATURES" or
somesuch that would be a list of identifiers indicating which features
were currently active? There would be no need to make assumptions -- if
"frames" were not in the list, an intelligent server would know not to
deliver a framed set of documents. With frames as a UA option,
everybody would be happy except for those who pay the bill to support
optional features. But even there, the bill-payer has the option to
avoid support for these features altogether.

David Perrell