Re: Netscape Plug-ins and <OBJECT>

Jukka Korpela (jkorpela@cc.hut.fi)
Fri, 19 Sep 1997 14:33:27 +0300 (EET DST)


Date: Fri, 19 Sep 1997 14:33:27 +0300 (EET DST)
From: Jukka Korpela <jkorpela@cc.hut.fi>
To: www-html@w3.org
In-Reply-To: <Pine.WNT.3.95.970917092232.-25877E-100000@hazel.hpl.hp.com>
Message-ID: <Pine.OSF.3.96.970919141157.27801B-100000@torvi.hut.fi>
Subject: Re: Netscape Plug-ins and <OBJECT>

There seems to be a lot of confusion around the OBJECT element.                 
(See http://www.hut.fi/home/jkorpela/HTML4.0/comments.html#obj )                
I hope things will be clarified, since OBJECT is really one of                  
the most important things in the HTML 4.0 draft.        

The current thread seems to deal with the association of various
processing methods (such as "plug-ins") with different objects.
This is an important issue too, and I think it supports my idea
of _requiring_ the type attribute. Notice that the issue
naturally divides into two sub-issues: the recognition of a
content type (media type) by a user agent, and the mapping
of content types to various operations. The latter is a browser
issue, not an HTML issue.

On Wed, 17 Sep 1997, Dave Raggett wrote:

> Assuming the browser has a dictionary that binds Content types 
> to plugins, then in principle, the following would work:
> 
>   <object data=foo>
>    a picture of foo
>   </object>
> 
> The above gets the Content type when it downloads the data.

Does it? As far as I can see, the data attribute may refer to any URL,
not just an HTTP URL. Even if the user agent gets a content type from
a server, wouldn't it be good to check that the type matches the one
expected by the author of the document? (In the case of mismatch,
there are various sensible things that a browser might do.)

So I think that the type attribute of OBJECT should be obligatory.
(There is little sense in including an object if the author does not
know its media type.)

>   <object data=foo width=100 height=60>
>    a picture of foo
>   </object>
> 
> This specifies the with and height for the image. The image
> will be scaled if necessary to this size.

Will it? The HTML 4.0 draft says, at
http://www.w3.org/TR/WD-html40-970708/struct/includes.html#visual
that the (deprecated) width and height attributes "give user agents an
idea of the size of an image or object so that they may reserve space
for it". It also says that "user agents may scale objects and images to
match these values if appropriate". This is confusing. (For example,
assume that I use an image on my page, using the width and height
attributes for the purpose of speeding things up, which is _the_ purpose
according to HTML 3.2 and still seems to be the _primary_ purpose
according to the HTML 4.0 draft. Now suppose the owner of the image
replaces it with a new version with slightly different dimension.
Some browsers would use the actual dimensions of the image whereas some
others would scale it, perhaps distorting the image badly if the
width:height proportions change.)

I think attributes which are to be deprecated should stay as they are.
HTML specifications should not encourage authors to _rely_ on rescaling.
Obviously the current wording of the draft can easily be read that way.

Yucca, http://www.hut.fi/home/jkorpela/