| Summary: | JS files served to XHR fail when you update the server, because they are cached | ||
|---|---|---|---|
| Product: | [ECD] Orion | Reporter: | John J. Barton <johnjbarton> |
| Component: | Server | Assignee: | Simon Kaegi <simon_kaegi> |
| Status: | RESOLVED WORKSFORME | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | malgorzata.tomczyk, simon_kaegi, Szymon.Brandys |
| Version: | 0.4 | ||
| Target Milestone: | 0.4 RC3 | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
|
Description
John J. Barton
We are supposed to have Cache-Control: no-cache headers on all dynamic content coming back from the server. It is possible we have missed some cases so we'll need to narrow down which XHR resources are being cached. BTW, no-cache means "cache but revalidate before use". As I understand the webkit bug, the late XHRs are not revalidated. But I could be wrong. See also http://html5boilerplate.com/docs/Version-Control-with-Cachebusting/ http://requirejs.org/docs/api.html 'urlArgs' http://bugs.dojotoolkit.org/ticket/13099 (In reply to comment #2) > BTW, no-cache means "cache but revalidate before use". As I understand the > webkit bug, the late XHRs are not revalidated. But I could be wrong. I modified the test case on webkit bug 30862 to verify that files served with Cache-control: no-cache do *not* exhibit 30862. Gosia did a change to make JSON responses no store -- leaving this in RC2 to verify. Tested. Closing as works for me but I'm sure we will need to continue watching and understanding how the browsers cache. |