[webkit-dev] Fwd: Debugging regression in Safari/WebKit Nightly Builds

[Dan Udey]

In regards to bug 5210:

Any ideas?

Begin forwarded message:

> From: Brothercake <cake at brothercake.com>
> Date: January 19, 2006 10:52:03 AM EST (CA)
> To: Dan Udey <dan at cdslash.net>
> Subject: Re: Debugging regression in Safari/WebKit Nightly Builds
>
> I think I've isolated the problem now - it appears that the  
> interpretor is overriding a global variable with a local variable  
> under specific circumstances.  I've setup a test page which  
> illustrates this happening, as well as showing a commented  
> variation of the same situation where it doesn't happen - http:// 
> www.udm4.com/webkit-debug.html
>
> By way of further experimentation - if the test code is enclosed in  
> another scope, and a new i variable is declared with "var i = 0"  
> then the code works as desired - it's a local var in a different  
> scope and therefore not affected.  But if the enclosing scope  
> continues to use the window variable (by simply going "i = 0") then  
> the problem is still seen.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.macosforge.org/pipermail/webkit-dev/attachments/20060119/f7f4a4ae/attachment.html