Bug 30788 – HistoryController does not handle some opaque URLs correctly

• Component: History
• Status: RESOLVED FIXED

[Alejandro G. Castro]

Created attachment 41890 [details]
Test reproducing the problem

This bug comes from the bug 30724, in the comment 2 it is explained the situation that is currently causing problems with the gtk+ DRT when testing opaque-base-url.html, but that it seems it could be reproduced with the test attached to this bug in other platforms. I've added the main authors of the opaque-base-url.html test to the CC, thanks in advance for the help.

The test tries to reproduce the error in other platforms, not just gtk+, trying to do a similar action done in the gtk+ DRT. Basically this kind of opaque urls are causing problems when the state is saved in the HistoryController, it is checked if that item is in current document and that tries to create a URL using an opaque URL that hits the checkEncodedString assertion about the first character in an url. I guess even this being an invalid URL (we should remove that from the opaque test then) we should handle the situation without reaching the assertion.

Can someone check the test with other platforms and see if the problem is there? I guess the patch should control the current document condition in case of this kind of opaque URLs, not sure if just avoiding saving or setting as current because there is not URL.

[Alejandro G. Castro]

(In reply to comment #0)
> Created an attachment (id=41890) [details]
> Test reproducing the problem
> 
> This bug comes from the bug 30724, in the comment 2 it is explained the
> situation that is currently causing problems with the gtk+ DRT when testing
> opaque-base-url.html, but that it seems it could be reproduced with the test
> attached to this bug in other platforms. I've added the main authors of the
> opaque-base-url.html test to the CC, thanks in advance for the help.
>

The html code is this, I'll try to check the assertion:

<script>
if (window.layoutTestController) {
  layoutTestController.dumpAsText();
  layoutTestController.queueReload();
}
</script>

<iframe frameborder=0 src="data:text/html,<iframe frameborder=0 src='<p>FAILURE</p>' width=10 height=10></iframe>"></iframe>

[Alejandro G. Castro]

The test is nowadays passing, closing the bug.