Comments on: Better code downloading with AJAX

By: Erik Arvidsson

Erik Arvidsson — Wed, 12 Oct 2005 17:14:33 +0000

I have looked at the code for Mozilla but like you I haven't really found where the bug is. I assume the bug is that they do not add the if-modified-since header and therefore the files will always be redownloaded from the server.

It is pretty easy to see what goes on in Mozilla using Live HTTP Headers. Then you'll see that it totally ignores last-modified.

By: Alexander Kirk

Alexander Kirk — Tue, 11 Oct 2005 07:34:29 +0000

I haven't yet had time to approve Erik's comment myself (by trying it out) but it did some research on the web which seemed to approve it.

Furthermore I had a look at the Source code part where a comment says:
1581 // Bypass the network cache in cases where it makes no sense: 1582 // 1) Multipart responses are very large and would likely be doomed by the 1583 // cache once they grow too large, so they are not worth caching. 1584 // 2) POST responses are always unique, and we provide no API that would 1585 // allow our consumers to specify a "cache key" to access old POST 1586 // responses, so they are not worth caching.

So, when using a GET request for loading data via XHR it should indeed be cached. I will verfiy this.

By: Michael Mahemoff

Michael Mahemoff — Tue, 11 Oct 2005 00:50:14 +0000

Great stuff! I need to add some discussion of caching in On-Demand Javascript.

I was interested in Eric's comment wrt XHR caching and Firefox and found this link - https://bugzilla.mozilla.org/show_bug.cgi?id=268844 - which confirms there was a bug,

And the commentary there links to an interesting comment in the WHATWG webapp spec:
"In particular, UAs must not automatically set the Cache-Control or Pragma headers to defeat caching." http://whatwg.org/specs/web-apps/current-work/#setrequestheader