Overview Description: image.setAttribute("align", "bottom") does not seem to conform to the HTML 4.0 transitional spec. Steps to Reproduce: 1) launch the url (i will add it upon completion of this bug) Actual Results: 1) if a line of text appears in red, it means that the property is different from what i tried to set it to. Expected Results: I set "img.align = bottom". A user might want to check to see that it is still set to bottom, but will be disappointed to see that it was changed (even though the behavior of bottom and baseline might be the same. Also, i would expect that invalid values would not be set, like "nothingatall". Build Date & Platform: 1999-05-11-08 build on Linux 5.2 (Apprunner) 1999-05-11-08 build on WinNT 4.0 sp4 (Apprunner) 1999-05-11-08 build on MacOS 8.51 (Apprunner)

added url.

Still waiting for a new test case for this bug.

sorry i let this slip through the cracks. the new testcase is posted to the same url above.

didn't realize this was assigned to me. reassigning to vidur.

Currently, we crash when the first setAttribute() takes place. Similar to bug# 10007. The image is not initially a floater. When the call to setAttribute goes through, we first reresolve style and then try to recreate frames. To recreate the frame for the image, we first delete the old one and insert a new one. When we're deleting, the style context seems to indicate that we're a floater (since we just reresolved style), even though the old frame wasn't. As a result, the frame tree is left in a weird state and we die on insertion of the new frame. *** This bug has been marked as a duplicate of 10007 ***

sounds good to me. marking verified.