Bug 3400 – setting the .src of an iframe to the same value does not reload page

• Component: DOM
• Status: RESOLVED FIXED

[Mark McCray]

When setting the .src of an iframe using .src = url, safari will not reload the iframe if the current url is the 
same as the new url.
However, using contentWindow.location.href and setting it to the same value to force a reload will work.
Is this correct behaviour?

The link above shows a page with two buttons to reload the frame using both methods listed above.
The default link that the page should reload a page that shows the current time with seconds. Clicking on 
the "src" button will load the page initially, but not refresh after that. Clicking on the "contentWindow" 
button will continually reload the iframe.
Thanks!

[Alexey Proskuryakov]

Created attachment 6349 [details]
proposed fix

I don't see how to make an automated test for this.

[Eric Seidel (no email)]

Comment on attachment 6349 [details]
proposed fix

We could extend DumpRenderTree to allow for an exposed JavaScript object which is shared between page loads within a test.

[Darin Adler]

Comment on attachment 6349 [details]
proposed fix

It's clear that this change will affect only the handling of the location and src attributes of frame and iframe objects. That's good, so I think the change is fine. I think that we might also want to rename the updateForNewURL function to clarify that it can be used even when the URL is not changing.

It should be straightforward to make a test for this, using the onload handler for the frame element. We'd want to test both location and src for both frames and iframes. Setting review- just because we need the test.

[Alexey Proskuryakov]

Created attachment 6369 [details]
proposed fix

Making tests wasn't exactly straightforward because of bugs like 7161, 3609 and 3580, but here they are :)

I couldn't come up with a new name for updateForNewURL(), but getting rid of this function by moving its code into setLocation() appears to be a better thing to do anyway.

[Darin Adler]

Comment on attachment 6369 [details]
proposed fix

Looks fine. r=me

I'd like the test case slightly more if it was a dumpAsText one.

[Alexey Proskuryakov]

The IFrame tests use dumpAsText. As for the other two tests- frame text isn't dumped by DumpRenderTree, so instead of hacking around it I just did what other fast/frames test do...