RE: Automatic loading of pages

On Sun, 26 Apr 1998, Jason White wrote:

(regarding the META HTTP-EQUIV REFRESH hack):

> In answer to Gregg's question, I believe the timed change of pages is
> supposed to happen whenever a document bearing this markup is loaded.

First, a little pedantry, if I may.  This is not non-standard HTML, as
someone called it earlier in the thread; in fact this is a standard
HTML way of incorporating an HTTP header into HTML markup.  What is
non-standard here is the HTTP header that this purports to supply.
There is no such header in the HTTP standard (rfc2068).

So, this is a proprietary feature, to which browsers will respond, or
not, or offer options, according to their design.

By the way, one _can_ send this header from the server, as a genuine
HTTP header, with any kind of document (not only HTML).  But when I
say "one can", I mean "it is technically feasible".  Whether it's a
good idea is a separate question.

> Do
> today's browsers offer a means of deactivating this feature? 

Some do, some don't.  And some recognise the header and politely
inform the reader about it without taking any automatic action (ex:
NCSA WinMosaic 3, Lynx).

> If not, then
> I would suggest (1) that the page authoring guidelines discourage its use; 

I agree that many of the uses to which it is put are inappropriate and
should be discouraged.  There are situations where it plays a useful
role.  To take just one primitive example, I have a network
performance page whose contents are updated every 10 minutes.  With
this proprietary hack, I can have the browser displaying the new data
automatically, without user action.  The same would apply for a
news-headlines page etc. 

A better-engineered solution would be a browser option whereby the
reader would request the page be automatically updated whenever it
expires. That would use only _standard_ features of the protocols, and
would be user-oriented.  It's sad that the big vendors were designing
their browsers for the authors, and not for the browser users.

It may be that use of the Refresh hack is already, in effect,
discouraged by the general mandate to use only standard features,
(though I'm afraid some authors act as if the "standard" was whatever
their favourite vendor implements). I'm not sure exactly what wording
is best for the guidelines, but I would not want to rule this out
entirely, as I think there are valid usages, albeit they are only a
fraction of the places where the hack is currently seen in use
("splash pages" and such nonsense inherited from notions of "the web
as tv"). 

So it would be a very good idea to encourage browser makers to
implement a preferences option, regardless of what the consensus comes
up with regarding advice to authors.

Received on Sunday, 26 April 1998 06:42:26 UTC