On Nov 24, 8:01 pm, Boris Zbarsky <[EMAIL PROTECTED]> wrote: > [EMAIL PROTECTED] wrote: > > So the problem must be in my code not a > > regression with the upgrade, but the upgrade appears to have broken my > > formerly working script. > > That would be a bug; point releases shouldn't be breaking web-facing > functionality (unless it's depending on a security hole, of course). > > > > > What I found was that 2.0.0.17 and .18 both on XP will work properly. > ... > > With version 2.0.0.18 on Vista a number of readyStatus == 3 are shown > > but a readyState == 4 is never reached. > > That's quite odd. I'm not aware of anything we do differently there > based on the Windows version.
More to report: I uninstalled 2.0.0.18 Vista, reinstalled 2.0.0.17 and proved that that version could run both the IFRAME load and the XHR. Then I used the Upgrade menu selection to install 2.0.0.18 on the Vista machine. I double-checked the user.js file and found that the syntax was that from the 2006 Ruderman document on the FF website. I had been experimenting with it and had asked him about the problem I had been seeing. So I made the user.js on both the XP and the Vista identical. Then I restarted 2.0.0.18 on the Vista and found nothing had changed: I could do the IFRAME load with those user_prefs but not the XHR. In fact, the Vista version does not even do same-site XHR because it never reaches a readyStatus of 4. Conversely, running on XP 2.0.0.18 has no problems. It looks increasingly like it is Vista that has changed. It might have been in an automatic update around the time of the release of 2.0.0.18. Hasn't anyone else reported a problem doing XHR even with the same origin? _______________________________________________ dev-security mailing list dev-security@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-security
