Timeing issue with FireFox dojo.browser.undo.addToHistory/dojo.io.bind changeUrl:

[ialpert@…]

Having a hard time reproducing this one in a test.

In a symptoms are changeUrl is enabled on a dojo.io.bind call i get two things

1. the browser instead of with changeUrl: "foo,bar,baz" getting somethign like

​http://localhost/blah/lastCurrentAddressBar.do#foo,bar,baz i get ​http://localhost/blah/#foo,bar,baz

2. the back call on the initial state seems to get triggered.

[Izaak Alpert]

this still doesn't seem to illicit the not using the current address, but with the debug lines in the browser.js there's sufficient time to cause the changeLocation to be called.

[James Burke]

I've noticed an issue on tests/io/test_BrowserIO.html where the back handler seems to get triggered. I think the core issue is that test page calls dojo.io.bind() as part of the dojo.addOnLoad() call. I don't think this should be done. I think just dojo.undo.browser.setInitialState() should be called as part of dojo.addOnLoad(), then user actions that trigger bind calls can register with dojo.undo.browser.

Could that be the issue you are encountering? calling a dojo.io.bind() with a back: function parameter as well as calling dojo.undo.browser.setInitialState() inside the page's dojo.addOnLoad() callback?

[Izaak Alpert]

I think this is exhibited regardless of the presence of a bind call in the addOnLoad(). None of the bind calls i've used implement a back: function (only changeUrl). I think this test works with or without the bind call, however i have to modify the original source (added some dojo.debug calls -- presumably slowing it down) to exhibit this behaviour

[Izaak Alpert]

Ok this was caused because i was setting the base tag of the document -- sorry

[James Burke]

Thanks for the more information. I'll close the bug now.