Bug 14709

Summary:

document.write in onload fails

Product:

WebKit

Reporter:

David Smith <catfish.man>

Component:

DOM

Assignee:

Nobody <webkit-unassigned>

Status:

RESOLVED DUPLICATE

Severity:

Normal

CC:

ap, aroben, mrowe, sornalatha.rathnasamy, vincent.bradley

Priority:

Version:

523.x (Safari 3)

Hardware:

Mac

OS:

OS X 10.4

URL:

http://dscoder.com/documentwritetestcase.html

Attachments:

Description	Flags
close the write content when the doc is closed	mrowe: review-
close the stream after write by appending </html> for the corresponding <html> tag added in write()	sam: review-
test case	mrowe: review-

David Smith

Reported 2007-07-22 01:43:00 PDT

The linked url shows "success" on the page in firefox, but is blank in webkit. It uses document.write in onload on the <body>. In addition, having the document.write in onload will also cause a subsequent document.write to fail. For example <body onload="document.write('success')"><script type="text/javascript">document.write('success');</script></body> will print nothing, but if the onload on the body is removed, it will print success.

Attachments
close the write content when the doc is closed (1.46 KB, patch) 2007-08-02 11:24 PDT, Sornalatha Rathnasamy	mrowe: review-	Details Formatted Diff Diff
close the stream after write by appending </html> for the corresponding <html> tag added in write() (2.90 KB, patch) 2007-08-14 11:57 PDT, Sornalatha Rathnasamy	sam: review-	Details Formatted Diff Diff
test case (413 bytes, text/html) 2007-08-14 13:29 PDT, Sornalatha Rathnasamy	mrowe: review-	Details
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.

Adam Roben (:aroben)

Comment 1 2007-07-22 01:46:21 PDT

The second example you describe shows a document.write happening *previous* to the one in the onload handler. The body of the <script> element will be executed as soon as it is parsed, which will happen before the onload event fires. My guess is that the document.write inside the <script> executes correctly, but then the document is cleared when the document.write in the onload handler does an implicit document.open. But this is just speculation.

Sornalatha Rathnasamy

Comment 2 2007-08-02 11:24:28 PDT

Created attachment 15809 [details] close the write content when the doc is closed

Sornalatha Rathnasamy

Comment 3 2007-08-02 11:25:15 PDT

(In reply to comment #2) > Created an attachment (id=15809) [edit] > close the write content when the doc is closed http://bugs.webkit.org/show_bug.cgi?id=14821 is a duplicate of this bug.

Darin Adler

Comment 4 2007-08-02 18:48:53 PDT

*** Bug 14821 has been marked as a duplicate of this bug. ***

Darin Adler

Comment 5 2007-08-02 18:49:29 PDT

Comment on attachment 15809 [details] close the write content when the doc is closed This patch is not for TOT. Must be for the S60 branch.

Mark Rowe (bdash)

Comment 6 2007-08-03 01:46:25 PDT

Comment on attachment 15809 [details] close the write content when the doc is closed I'm going to mark this r- as the patch is for the S60 branch while the bug was reported against trunk. This means that the patch does not address the bug that was reported.

Sornalatha Rathnasamy

Comment 7 2007-08-03 05:51:43 PDT

Comment on attachment 15809 [details] close the write content when the doc is closed this patch is applicable to tiger branch as well not just S60. Could someone please review ?

Sornalatha Rathnasamy

Comment 8 2007-08-03 05:51:49 PDT

Comment on attachment 15809 [details] close the write content when the doc is closed this patch is applicable to tiger branch as well not just S60. Could someone please review ?

Mark Rowe (bdash)

Comment 9 2007-08-06 02:38:20 PDT

Comment on attachment 15809 [details] close the write content when the doc is closed As mentioned, the bug is reported against trunk so a patch for the Tiger branch does not resolve the bug.

Alexey Proskuryakov

Comment 10 2007-08-06 04:23:46 PDT

Also, the patch for TOT should include a test case and a ChangeLog entry. Please see <http://webkit.org/coding/contributing.html> for information on contributing code.

Sornalatha Rathnasamy

Comment 11 2007-08-14 11:57:39 PDT

Created attachment 15964 [details] close the stream after write by appending </html> for the corresponding <html> tag added in write()

Alexey Proskuryakov

Comment 12 2007-08-14 13:22:51 PDT

The patch still doesn't have a test, so it cannot be landed. However, I'm leaving the review flag on for someone to review code changes. BTW, this seems to be a duplicate of bug 8961.

Sornalatha Rathnasamy

Comment 13 2007-08-14 13:29:23 PDT

Created attachment 15968 [details] test case test case

Mark Rowe (bdash)

Comment 14 2007-08-17 23:26:06 PDT

Comment on attachment 15968 [details] test case Your test case should be included in the patch, along with the expected results. It seems that this test should also be a text-only test rather than dumping the rendertree.

Mark Rowe (bdash)

Comment 15 2007-08-17 23:36:16 PDT

Comment on attachment 15964 [details] close the stream after write by appending </html> for the corresponding <html> tag added in write() Your ChangeLog entries should include information about the change as well as a reference to the bug is being fixed, as well as your name rather than "webcore" in the header line. The name "m_beginWriteStream" seems less than ideal. It reads more like a function name due to it starting with a verb. I can't suggest a better name as I'm not entirely clear on what the purpose of the variable is.

Sam Weinig

Comment 16 2007-08-19 17:38:25 PDT

Comment on attachment 15964 [details] close the stream after write by appending </html> for the corresponding <html> tag added in write() As noted above, this needs a more detailed ChangeLog and an included test case. r-

Maciej Stachowiak

Comment 17 2007-08-19 21:43:03 PDT

I'm not sure the basic approach here is right: a) Writing a </html> close tag shouldn't be necessary to have the written contents actually appear; the fact that it is likely indicates some deeper problem. b) This probably means that certain kinds of document.write *after* onload won't work right either, and there isn't code around to write the </html> in such cases.

Alexey Proskuryakov

Comment 18 2007-09-26 21:27:46 PDT

This bug doesn't seem active anymore; actually closing as a duplicate. *** This bug has been marked as a duplicate of 8961 ***

Note You need to log in before you can comment on or make changes to this bug.