Re: Frames spec & the NOFRAMES tag

nir dagan (dagan@upf.es)
Wed, 4 Mar 98 12:09:32 MET


From: nir dagan <dagan@upf.es>
Message-Id: <199803041109.MAA28232@sahara.upf.es>
To: bhaller@apple.com
Date: Wed, 4 Mar 98 12:09:32 MET
Cc: www-html@w3.org
In-Reply-To: <9803040030.AA10532@santafe.apple.com>; from "Ben Haller" at Mar 3, 98 7:39 pm
Subject: Re: Frames spec & the NOFRAMES tag

I think that the problems in the spec. for frames arise
from the fact that frames were implemented before there was a spec.

The NOFRAMES element was a content of FRAMESET by the recommendation of 
the implementors.

This makes it logically impossible to have a BODY element in it.

However its not clear why the spec does not allow the attributes
that go in BODY for the NOFRAMES element.

If frames were to designed from scratch the frameset document 
would have had three parts: HEAD, FRAMESET, and BODY.

Another problem with the implementation of frames is the fact that
it was "valid" in Netscape's implementation to have only a cols or
rows attribute in FRAMESET, but not both. This is inconsistent with the 
restriction from SGML that does not allow to condition the presence 
of an attribute by the absence of another. 
What Netscape should have done in to have two different
elements FRAMESETROWS and FRAMESETCOLS. 

Now the spec allows both together. This implies that frames can become
non-rectangular by resizing them. 

Assume you have a grid of 2x2:
then pull down the  upper border of the frame in the lower left, and 
then pull to the right the right border of the same frame.

Lets see if this will be implemented...

Best regards,

Nir Dagan.
email: dagan@upf.es
URL: http://www.econ.upf.es/%7Edagan/