RESOLVED FIXED6385
REGRESSION: Unnamed frames return a generated name as window.frameElement.name 
https://bugs.webkit.org/show_bug.cgi?id=6385
Summary REGRESSION: Unnamed frames return a generated name as window.frameElement.name
Alexey Proskuryakov
Reported 2006-01-05 10:56:08 PST
This is a rather old regression (at least from September).
Attachments
test case (2.71 KB, application/zip)
2006-01-05 10:56 PST, Alexey Proskuryakov 		no flags
Details
patch that fixes this regression (3.30 KB, patch)
2006-01-22 23:17 PST, Darin Adler 		mjs: review+
DetailsFormatted DiffDiff
correct the test case (2.37 KB, patch)
2006-01-23 21:56 PST, Alexey Proskuryakov 		darin: review+
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Alexey Proskuryakov
Comment 1 2006-01-05 10:56:58 PST
Created attachment 5501 [details] test case I wonder if it's possible to create self-contained tests for such issues...
Alexey Proskuryakov
Comment 2 2006-01-05 12:25:02 PST
Ah, the actual returned name is "<!--framePath //<!--frame0-->-->", which gets displayed as "-->". Firefox returns an empty name, not a generated one like this... Interestingly, HTMLElement::frameGetter is not invoked to get the name at all, although it has code for FrameName case; Window::getValueProperty is invoked, and retrieves the name as m_part->name().
Darin Adler
Comment 3 2006-01-06 11:17:03 PST
Are you sure this is a regression? We've always generated names like this internally.
Alexey Proskuryakov
Comment 4 2006-01-06 12:58:27 PST
(In reply to comment #3) Yes; with stock Safari 2.0.2, window.frameElement.name returns the original empty name (I've double- checked using escape(), to make sure the returned name wasn't just invisible because of its comment-like form).
Alice Liu
Comment 5 2006-01-11 17:31:08 PST
<rdar://problem/4400230>
Joost de Valk (AlthA)
Comment 6 2006-01-22 04:36:53 PST
Adding Regression keyword.
Darin Adler
Comment 7 2006-01-22 23:17:28 PST
Created attachment 5865 [details] patch that fixes this regression
Darin Adler
Comment 8 2006-01-22 23:18:06 PST
The regression is fixed, but window.name still gives the strange frame name. We might not want that and it's worth doing some additional testing and perhaps filing additional bugs.
Maciej Stachowiak
Comment 9 2006-01-23 13:41:15 PST
Perhaps generated names should be kept as a purely internal thing. I'm not really sure what we use them for.
Alexey Proskuryakov
Comment 10 2006-01-23 21:55:48 PST
Filed bug 6751 for the window.name issue. Here, the test case lacks "resources/" for the left frame, reopening because of this. Also, perhaps it would be better to use escape() on the names, to make sure the right one is really empty.
Alexey Proskuryakov
Comment 11 2006-01-23 21:56:18 PST
Created attachment 5901 [details] correct the test case
Darin Adler
Comment 12 2006-01-23 23:18:28 PST
Comment on attachment 5901 [details] correct the test case You're welcome to update to use escape() -- I did the other half.
Darin Adler
Comment 13 2006-01-23 23:18:48 PST
Remaining bit is not P1.
Eric Seidel (no email)
Comment 14 2006-01-31 21:20:41 PST
Removing Regression keyword from bugs already fixed.
Note You need to log in before you can comment on or make changes to this bug.