[Webkit-unassigned] [Bug 31446] crash in chromium when loading multipart/x-mixed-replace data

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=31446





--- Comment #13 from Tony Chang (Google) <tony at chromium.org>  2009-12-02 11:06:51 PST ---
(In reply to comment #12)
> Actually, doesn't the FrameLoaderClient get notified during the call to
> didReceiveResponse?  In that case, would the solution I just proposed really
> work?

It works in that it doesn't crash, but it doesn't render anything if the
content is less than 1024 bytes.  Using the suggested change (wait for data
before sending the header), here's a video when the server sends 512 bytes of
data:
http://ponderer.org/tests/data512.ogg
http://ponderer.org/tests/multipart-512.py

Here's a video when the server sends > 1024 bytes of data:
http://ponderer.org/tests/data1024.ogg
http://ponderer.org/tests/multipart-1024.py

The ToT Chrome handles the above two cases correctly, but the renderer crashes
when navigating away (not really noticeable unless you try to navigate to the
same domain).

I think sending the headers right away tells the doc loader than it can render
the data it has so far.  This allows data less than 1024 bytes to render
immediately.  One option would be to have Chrome match CFNetwork and try to
make the document render partial data.

Firefox handles both cases properly and Safari doesn't handle the 512 case
because of CFNetwork merging two parts into a single document.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.