Bug 43371 - REGRESSION (HTML5 tokenizer): bush-whacker.com stops rendering halfway
Summary: REGRESSION (HTML5 tokenizer): bush-whacker.com stops rendering halfway
Status: RESOLVED WORKSFORME
Alias: None
Product: WebKit
Classification: Unclassified
Component: Layout and Rendering (show other bugs)
Version: 528+ (Nightly build)
Hardware: Mac OS X 10.5
: P1 Major
Assignee: Nobody
URL: http://www.bush-whacker.com/
Keywords: Regression
Depends on:
Blocks: 41115
  Show dependency treegraph
 
Reported: 2010-08-02 12:35 PDT by Kevin M. Dean
Modified: 2021-11-03 03:13 PDT (History)
3 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Kevin M. Dean 2010-08-02 12:35:20 PDT
With Safari 5.0.1 under PPC/intel and 10.5/10.6 the pages of this site do not render past the photo header.

Up to the day of Safari 5.0.1's release, the latest Webkit releases were rendering this site correctly. I was viewing the site on the day of 5.0.1's release, but hadn't revisited the site with 5.0.1 until today. It seems that after Safari 5.0.1's release, the latest webkit releases and even past releases that worked under 5.0 no longer work until you get all the way back to Nightly r61173 (mid-June). It does work in Safari 5.0.1 itself though just not when running the latest webkit after installing 5.0.1.

I remember there were some other rendering issues between r61173-r61351 so maybe this is related.
Comment 1 Alexey Proskuryakov 2010-08-04 06:21:49 PDT
I don't think this is related to Safari application in any way. I can reproduce this with Safari 5.0 and ToT WeKit just as well.

This seems to be another example of incompatible change in HTML5, which no longer treats "<tag<tag>" the way HTML4 did (according to a W3C validator warning message). But whatever the specs say, this is definitely an incompatible change for WebKit.

The offending snippet is:

Company<br /><br						</div>
Comment 2 Adam Barth 2010-08-04 09:44:17 PDT
Sad face.
Comment 3 Adam Barth 2010-08-04 09:45:56 PDT
As usual, the page has the same problem in IE.
Comment 4 Adam Barth 2010-08-04 09:53:18 PDT
I have some data on this issue, but the way.  This pattern (the missing >) is not that uncommon.  The problem is that some of the pages want to be tokenized the old way and some want to be tokenized the new way.  Some folks from Opera said they've tried both ways and sites break either way.  Of course, we only get reports of the new breakage because of selection bias in folks running the nightly.
Comment 5 Kevin M. Dean 2010-08-04 11:35:16 PDT
So, it looks like the root of the problem was some extra html and a form that was being placed in their carousel banner under the Smoker section. I had it removed and it appears to have fixed the rendering issue. It seems that some of the html was being trimmed because it was too large for the area and it happened to trim the missing > off the <br> and contribute to the non-rendering issue.
Comment 6 Adam Barth 2010-08-04 15:33:44 PDT
Thanks for fixing the site.  There's another bug (actually Bug 40961) that's tracking this issue in general.
Comment 7 Alexey Proskuryakov 2010-08-05 04:54:29 PDT
> sites break either way

If it's true that shipping Gecko and WebKit behavior matches HTML4, and both behaviors cause site breakage, then it's really surprising that HTML5 mandates a change. How is that an improvement?
Comment 8 Adam Barth 2010-08-05 09:41:51 PDT
Because IE ships the other way.  Here's the spec bug:

http://www.w3.org/Bugs/Public/show_bug.cgi?id=9985

We can escalate to an issue, but without Mozilla's support, I suspect we won't be able to change the spec.
Comment 9 Alexey Proskuryakov 2010-08-05 11:45:50 PDT
I agree that we'd surely need Mozilla support to reverse the decision.