Bug 9370

Summary: Nothing rendered when missing ending </script> tag
Product: WebKit Reporter: Alex Shanks <ashanks>
Component: DOMAssignee: Nobody <webkit-unassigned>
Status: RESOLVED INVALID    
Severity: Normal CC: ashanks, ddkilzer, ian, joost, nickshanks
Priority: P2    
Version: 417.x   
Hardware: Mac (Intel)   
OS: OS X 10.4   
URL: http://www.host41.com/
Attachments:
Description Flags
Webarchive of original revision for the website
none
Original source (No </script> in <head> javascript) none

Alex Shanks
Reported 2006-06-08 21:11:58 PDT
When trying to access www.host41.com, nothing is rendered - page is blank (except for the <title> of the page, at the top of the window). When you go to "View Source" in Safari, the HTML markup is there. In other browsers (mainly Gecko-based ones) the website loads fine. Someone on IRC noted that it is missing a </script> tag, and that is probably why, but I thought maybe this bug report would help maintain compatibility with Mozilla-based browsers.
Attachments
Webarchive of original revision for the website (3.70 KB, application/octet-stream)
2006-06-08 21:19 PDT, Alex Shanks 		no flags
Details
Original source (No </script> in <head> javascript) (4.31 KB, text/html)
2006-06-09 12:23 PDT, Alex Shanks 		no flags
Details
View All Add attachment proposed patch, testcase, etc.
Alex Shanks
Comment 1 2006-06-08 21:19:20 PDT
Created attachment 8779 [details] Webarchive of original revision for the website The webmaster has changed the site to fix a Javascript error, that caused it to not render correctly. I'm attaching a webarchive of the original version. You can keep the bug open if you still want to use it to help maintain Gecko compatibility.
David Kilzer (:ddkilzer)
Comment 2 2006-06-09 01:27:05 PDT
See also Bug 3905 and Bug 6314.
Joost de Valk (AlthA)
Comment 3 2006-06-09 01:29:35 PDT
I'm inclined to go even further, i'd guess that any unclosed tag in the head SHOULD be closed on head close, amd that the three bugs about unclosed <style>, <script> and <title> are in fact dupes of the same problem.
David Kilzer (:ddkilzer)
Comment 4 2006-06-09 02:10:26 PDT
(In reply to comment #3) > I'm inclined to go even further, i'd guess that any unclosed tag in the head > SHOULD be closed on head close, amd that the three bugs about unclosed <style>, > <script> and <title> are in fact dupes of the same problem. On the surface it looks that way, but it's a little more complex than this. Each tag has its own handling code and thus must be special-cased. Also when the tag appears in the <head>, this is a different code path than when the tag appears in the <body>. I got stuck a couple months ago on Bug 3905 and Bug 6314 when someone found a page in production that cause the tokenizer (and parser?) to "run out" of HTML to process while a page was loading, but only because the network couldn't feed the tokenizer/parser fast enough. I need a reproducible test case for this condition before I can fix the bug, and it's going to have to be an http test. I was also hoping that not looking at the tokenizer/parser code for a couple months would give me some fresh ideas when I went back to it.
David Kilzer (:ddkilzer)
Comment 5 2006-06-09 02:26:22 PDT
(In reply to comment #1) > Created an attachment (id=8779) [edit] > Webarchive of original revision for the website Alex, could you upload the HTML source of the original web site? The webarchive file doesn't contain the full, original HTML document.
Alex Shanks
Comment 6 2006-06-09 12:23:45 PDT
Created attachment 8787 [details] Original source (No </script> in <head> javascript) I didn't have a copy of the original source, but I talked with the webmaster and he said that he just moved the script to the bottom of <body> and added </script>.
David Kilzer (:ddkilzer)
Comment 7 2008-02-22 18:41:28 PST
What does HTML5 say about this?
Ian 'Hixie' Hickson
Comment 8 2008-02-22 19:51:46 PST
Script is only closed by </script> or EOF, for security reasons.
David Kilzer (:ddkilzer)
Comment 9 2008-02-22 20:31:07 PST
This matches HTML5 behavior. Closing as RESOLVED/INVALID. See Comment #8.
Note You need to log in before you can comment on or make changes to this bug.