Re: Frames sites.

I would like to explain in detail my controversial position 
regarding frames. My claim that frames are visual by 
nature is based on comparing  usability  
of a site with frames with a site without frames.

Consider the following simple example.

The site has a table of contents index.html and two 
"departments".

Without frames: The main page of the site is index.html, 
and it has links to both departments.
Each department has a link "home" to index.html and 
horizontal link to the other Department, given in
a navigation bar duplicated at the top and bottom of the page.

A disadvantage from the point of view of uasbility is 
that in order to change to another department, the user  has 
to go back to index.html or use the navigation bar, a feature that 
may require some scrolling if the page is not very short.

The solution with frames is to keep the navigation bar 
on a fixed location on the screen. Then the user simply
shifts the mouse to the navigation bar and follows the relevant link.
This may save him some time compared to the noframes alternative.

The mechanisms that render frames in a linear fashion reduce the 
frame structure to exactly the no frame structure. Thus, there is 
no usability *gain* for rendering frames lineary.
This is what I mean by "visual in nature." Frames have usability advantages
only if the user can have a direct multiple *view*.

Another problem is that framed sites usually do not have the navigation bar
on each page since it is assumed that the user uses frames. 
Then if these pages are used for the no-frames version 
the user must go back to index.html and cannot navigate horizontally.

A possible solution is to include this navigation bar in every page
and use the same pages for both the frames and no-frames versions.

The drawback of this method is that it may reduce usability when viewing
the site with frames. Having too many navigation bars may be confusing, 
as the user is overloaded with information.

Think here of a large site rather than the two departments example.
If a site has only 5 departments that are not "too deep",
frames reduce uasability anyway, since their complexity is a bigger burden
than the gain of the above mentioned mouse argument.

In this regard of large sites, the fixed navigation bar (separate frame) can connect 
the top level departments, and sub-depratments are connected by navigation 
bars on the pages themselves. 

Making a joint version reduces usability for one configuration.

Another serious problem of reusing the pages, is the fact that 
links in the framed version have target attributes defined.
If you use these pages without frames every link with a target 
opens in a new window. This effectively disables the "back" key
and forces the user to close/minimize windows all the time.

Opening new windows happens because the browser doesn't 
find a window with the name assigned by the target.
This practice of resolving targeting is mentioned  in the notes on 
implementation of the HTML4.0 spec.

So reusing the pages is possible in machines that don't open windows
e.g., Lynx for DOS, or in machines that allow to turn off opening windows 
e.g. Opera. But it is a usability disaster in Netscape 4 for example.
(more commom than all versions of Lynx and Opera tigether)

Therefore an author who uses frames to increase usability would
like to have a separate no-frames version for the very same purpose.
There is no escape. 

Best regards,
Nir.
 
Nir Dagan                            
Assistant Professor of Economics      
Universitat Pompeu Fabra
Barcelona (Spain)

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

Received on Thursday, 5 March 1998 16:49:21 UTC