When we do fragment-loading we should be storing the innerHTML of the #page fragment. This will prevent us from having to do another GET upon going back and forward in history.

Would this preserve the metrics logs for the server? And how long would we cache that for?

(In reply to Andy McKay [:andym] from comment #1) > Would this preserve the metrics logs for the server? > And how long would we cache that for? It's not cached. It's just in the history since that page load. So upon refreshing all of that is blown out and we start from scratch. But ordinarily when you browse sites that do full, synchronous page loads we don't make a roundtrip to the server when you click the back button. So if this is just adding parity with how pages usually work. Right now we're just making redundant requests (which ain't bad for inflating our traffic ;)). All of the webtrends calls are done upon "fragment load" so going back and forth in history would just work. I'm not sure what metrics looks at - so I'm uncertain there.

(In reply to Chris Van Wiemeersch [:cvan] from comment #2) > It's not cached. It's just in the history since that page load. So upon > refreshing all of that is blown out and we start from scratch. But > ordinarily when you browse sites that do full, synchronous page loads we > don't make a roundtrip to the server when you click the back button. So if > this is just adding parity with how pages usually work. Right now we're just > making redundant requests (which ain't bad for inflating our traffic ;)). Great. > All of the webtrends calls are done upon "fragment load" so going back and > forth in history would just work. I'm not sure what metrics looks at - so > I'm uncertain there. I think it's webtrends so thats good. /cc'ing daniel to double check.

This is conveniently fixed by the caching mechanism added in bug 758507.