Steps to reproduce: open the bug URL, mouse over the map. Expected results: cursor changes to a hand. ToT results: cursor changes to an icon of a missing image. Since shipped Safari doesn't support custom cursors at all (bug 5689), this problem doesn't occur, I'll file a separate P1/regression bug.
Created attachment 8401 [details] test case
CGImageSource already has support for .ico files, which have identical format (.cur differs by a single byte in the header). I have filed <rdar://4556355> for CG to support .cur. Keeping this bug open, too - the format is pretty simple, and we don't necessarily have to wait for CG to support it. Microsoft documentation: <http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnwui/html/msdn_icons.asp>.
Looks like .cur cursors are actually supported by CG, but not fully - some are rendered fine, some are rendered incorrectly, and others are just ignored.
This works in Safari for Windows on XP Pro (a four-directional arrow cursor appears instead of a hand over the map--is this the equivalent in Windows?). It may work in Mac OS X 10.5 Leopard as well.
That extra byte of difference between a .cur and a .ico includes the anchor point of the cursor; this information is currently being ignored, which results in bug 15779. As pointed out in comment #4, even the minimal support for showing the cursor doesn't work with Safari 3.1 on OS X 10.4; this is separately filed as bug 15783.
The test case passes for me with the latest nightly build r50918 under Mac OS 10.5.8. So, this issue seems to have been resolved. Alexey, are there still issue with .cur support? Can we close this bug? (In reply to comment #5) > That extra byte of difference between a .cur and a .ico includes the anchor > point of the cursor; this information is currently being ignored, which results > in bug 15779. > > As pointed out in comment #4, even the minimal support for showing the cursor > doesn't work with Safari 3.1 on OS X 10.4; this is separately filed as bug > 15783.
Partial support for .cur files is present in CG on Leopard - they are decoded successfully, but we don't get to know where hotspot is. So, adding support directly to WebKit is still an option. The actual page at bug URL still doesn't work correctly, so we need a new reduction.
Bug 30820 looks like it should have solved this case, but I can still reproduce on r51062. This is easy to reproduce with the Google Maps API: http://gmaps-utility-library-dev.googlecode.com/svn/trunk/geometrycontrols/examples/test.html Steps to reproduce: 1) Load demo 2) Click jagged line icon to toggle on polyline drawing 3) Observe that points are drawn at the upper-left corner of the cursor, not at the centre This works as expected in IE/FF (although FF for Mac has no custom cursor support at all AFAICT).
I can also confirm that custom .CUR cursors do not behave acceptably in WebKit r51881 due to bug 15779.
No comments on this bug for 1.5 years, and the issue now seems to be resolved.
It's not resolved on Tiger or Leopard (see the bug title), but it seems safe to assume that no one is going to work on that now.