"Anthony Jones" wrote:

> 
> "Rich-F" <RichF@d...> wrote in message
> news:EE64C55B-B026-46E7-8974-FE8FAAE1C0A4@m......
> > "Anthony Jones" wrote:
> >
> > >
> > > "Rich-F" <RichF@d...> wrote in message
> > > news:61C826B1-E1F2-4E6F-87FF-83C675E9FDF4@m......
> > > > "Anthony Jones" wrote:
> > > > <snip>
> > > >
> > > > > The point I was trying to communicate is that a browser offers the
> user
> > > the
> > > > > opportunity to override the normal operation of the local cache by
> using
> > > the
> > > > > refresh button.  This causes all requests to fulfill the page to
> pass
> > > > > through local cache to the origin server regardless of whether there
> is
> > > > > fresh instance of the resource in the cache.
> > > > >
> > > > > I'm trying to implement a similar feature for resources that have
> been
> > > > > fetched with XMLHTTP.  For performance reasons I do not want to send
> a
> > > > > request every time it's needed however I want to offer the user the
> same
> > > > > 'refresh' option.
> > > > <snip>
> > > >
> > > > Just considering this last point.....
> > > >
> > > > The only method that worked reliably for me was adding the
> > > If-Modified-Since
> > > > header and making sure that the date I used was definitely older than
> any
> > > > timestamp that might be associated with the file.  (I used IE6 and
> > > > XMLHTTP.3.0)
> > >
> > > Ok thanks for the input.  However since changing the If-Modified-Since
> in
> > > that way would always result in 200 response from the server and never a
> 304
> > > it's the same is simply using the the load method on a DOM with
> ForceResync
> > > enabled which is what I was hoping to avoid.
> >
> > (we're going round in circles here)
> 
> Yes we are
hmmmmm.
> 
> >
> > If you want to absolutely force the request to go to the server then set
> > If-Modified-Since to something really old.
> >
> > If you want to give the server a chance to reply with a 304 then set
> > If-Modified-Since to the 'real' modified date as per my previous code
> example.
> 
> Your previous code example knew what the 'real' modified date is from a
> previous fetch. As I pointed out in response to that; trapping, persisting
> and retrieving that piece of information is a complication way beyond the
> worth of the feature.

Sometimes, to achieve a simple result you have to put in a disproportionate 
amount of leg-work.

> I've also already stated that I'm not seeing the results in my testing that
> you appear to be.
> I've tried setting the If-Modified-Since header but it doesn't force a round
> trip to the server.

Works for me.

>This is as I would expect, the modification date of a resource has no bearing on the resources age in the cache.

The wording of that sentence suggests to me that you've misunderstood the 
effect/purpose of the header.

The overall tone of your responses indicates that you know enough to solve 
this by yourself, - the best of luck to you.