[ibeatty]

Greetings.

A web browser can actually send (at least) two different kinds of GET requests, and one instructs the server to flush caches and regenerate the content. (I forget the HTTP codes.) FireFox sends this "give me a fresh copy, dammit!" code when you use shift-command-(reload menu item). I think Safari might do it if you reload twice in a row, but I'm not 100% certain about that.

I discovered this when I ran into a caching problem that sounds similar to the one that started this thread. I was doing through-the-web authoring with the Plone content management system, and images I'd uploaded to replace others weren't displaying properly; the old image was still being sent (though at the new image's dimensions, which was a bit strange). I eventually convinced myself that it wasn't being cached on my local machine, by trying from multiple browsers and machines. But if I used FireFox's shift-command-reload, *bing* the server bypassed its own cache and the new image was displayed. I verified this interpretation by solving the same problem another way, going to the server's control panel and making it dump all its caches. I also looked at the server logs in detail, and noticed that a different request came in.

And that little feature makes Firefox very, very handy for authoring/development with servers that do any server-side caching, or when you have a caching proxy anywhere between you and the server. It would be a great feature to have in OmniWeb.

HTH,
:ian