Bug 32762 – upstream platform/chromium/plugins/get-url-with-blank-target.html

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Dirk Pranke]

Upstreaming a test from chromium.org - Chromium has chosen to implement slightly different behavior for LayoutTests/plugins/get-url-with-blank-target.html . See http://code.google.com/p/chromium/issues/detail?id=24186 for more context.

[Dirk Pranke]

Created attachment 45218 [details]
Patch

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 45218 [details] without any errors.

[Eric Seidel (no email)]

I'm confused by how this is different from the original and if this test still needs to be forked now that chromium is not secret.

[Dirk Pranke]

I'm not sure I fully understand how it's different, either, but according to http://code.google.com/p/chromium/issues/detail?id=24186 it is, and we don't want to merge the two.

[Eric Seidel (no email)]

That bug seems to suggest that WebKit's existing behavior, at least on the Mac, is wrong.  So I'm not sure that adding this test is the correct fix.

[Dirk Pranke]

Okay. I suggest then that we remove this test downstream, change this bug to suggest that the generic test is wrong (and WebKit Mac is wrong), and then either fix the test, fix the code, or then agree to disagree. Sound good?

[Eric Seidel (no email)]

The relevant comments from the Chromium bug:
I'm in this code, and ran into this bug.

I think the idea that webkit wants this window to not open is incorrect.

Reasons:
 1) The spec doesn't say so
 2) Firefox doesn't behave that way.  (It does open the window but does not
callback to the JS notify routine at all)
 3) Safari 3.1 for windows does not behave that way.  (It opens the
apple.com URL in a new window and returns error success).
mbelshe on March 31 2008 12:09
(Fixed / Fixed)
Notes
I fixed the layout test for this to reflect my latest comments.

[Eric Seidel (no email)]

The original test was written by Anders, and reviewed by aroben and darin.

[Eric Seidel (no email)]

Comment on attachment 45218 [details]
Patch

I don't think it makes sense to block upstreaming of Chromium's tests.  However, i do think it makes sense to have a bug about the differing behavior between this test and the original.  This bug could serve that purpose.

"behavior. i" seems to be a typo in the ChangeLog.

I think we should probably just land this as-is, and follow-up with a more informed fix to make Chromium and other WebKit ports agree here.

Lets fix the ChangeLog to note that this behavior difference should be remedied but for now we're at least testing the behavior.

+var NPERR_GENERIC_ERROR = 1;
line isn't really needed in this edited version of the test, but does no harm.

So I'll r+ this one, but since you're not a committer quiet yet, please post a new patch which updates the changelog to explain the intent to unify these behaviors.

[Dirk Pranke]

sounds good.

[Dirk Pranke]

Created attachment 45408 [details]
patch w/ revised ChangeLog entry

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 45408 [details] without any errors.

[WebKit Commit Bot]

Comment on attachment 45408 [details]
patch w/ revised ChangeLog entry

Clearing flags on attachment: 45408

Committed r52506: <http://trac.webkit.org/changeset/52506>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.