W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1995

reload discussion

From: Balint Nagy Endre <bne@bne.ind.eunet.hu>
Date: Wed, 16 Aug 1995 07:56:03 +0200 (MET DST)
Message-Id: <199508160556.HAA00508@bne.ind.eunet.hu>
To: Albert-Lunde@nwu.edu
Cc: http wg discussion <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
Lou Montulli and Albert Lundle discussing the reload operation:
[Albert]
> > > If the user requests a "Reload" operation, they are asking for a
> > > new copy of the resource from its origin.
[Lou]
> > We obviously have different opinions on what a "reload" should do.
> > 
> > My interpretation of "reload" is to check everything on the page
> > and retransfer any objects that have changed.
> 
> This is a case where it doesn't pay to be too clever.
> 
[Albert]
> If everything works perfectly, there is seldom a reason for
> people to use "reload". The two cases I use it are when I have
> a document that looks like it was corrupted by a transmission
> error or when I have outside information that a document *has*
> changed (usually because I'm editing it.)
> 
> In neither case do I want a local or remote cache be in the picture.
I use Lynx, Netscape and sometimes Mosaic.
All of them has a refresh function - e.g. screen refresh -
I quesst, normally this shouldn't involve any http action, using local cached
copy of the document.
I just now realised, that the "Pragma: no-local-cache" proposed by John Franks
can prevent this. <em> I dont want http actions to refresh my lynx screen when
it's garbled by talkd or write! </em>
The reload funtion, on contrary does start a http action, but which action it
really does, I not investigated till now.
My expectations to the "reload" function are:
re-check the status of the document, and reflect changes in case it's changed.
This should be done using HEAD method (HTTP/0.9 clients) or
GET/If-Modified-Since (HTTP/1.x clients).
If I am not satisfyed with the result - (I expected a new version, 
I'm sure that the document changed), I will request a reload again.
In this case, a good client may issue a plain GET, if it knows, that is
talking to the origin server directly, or a GET/no-cache if knows, that
there is some intermediate proxy. While pragmas have effect only on
intermediates (and I don't want changes in this), it's safe to add the no-cache
pragma, when no-intermediate proxies are involved in the action.
Can we agree on this approach?

Andrew. (Endre Balint Nagy) <bne@bne.ind.eunet.hu>
Received on Tuesday, 15 August 1995 23:00:47 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:25 EDT