[webkit-dev] How relevant is the data at the listed url?

[Mark Rowe]

On 20/05/2008, at 13:17, Scott Haneda wrote:

> I downloaded the nightly, and set the bash env var as per  
> instructions.
> `run-safari` does not seem to work, or I need a full path to it, but I
> can not locate the app. Has the procedure changed?
> Just launching Safari nightly from the Finder, and then running leaks
> safari does generate a report, telling me there are leaks. Should I
> send that in, or am I performing the steps wrong?
> I am trying to pin down a slowdown issue in Safari, where entering a
> URL and pressing return does nothing for several seconds. Eventually,
> action is taken and the page loader animation starts spinning. I use
> cmd-return often to inspire a new page to load in a new tab, the new
> tab is always made, but it is empty for several seconds, and then
> loads in.
> Quit Safari, and the behavior goes back to normal. Right now, I am
> testing to see if this is input manager related, so I have pulled out
> the one that I was using. I am running a meta refresh every one second
> on a local html file. Every page load uses 100k more memory, and
> growing.
> It is curious in that it is a local file, and the page is just
> refreshed, so I can not see why after the first page load, Safari
> needs more memory, it should be 100% cached.
> Thanks for any information in regards to this.

The data at which URL?  Based on some of the facts you mention, I  
suspect you're referring to <http://webkit.org/quality/ 
leakhunting.html>.  Those steps are correct and will work if  
followed.  Note that step 1 involves building a debug build of WebKit  
from source, not downloading a nightly build. The run-safari script is  
part of the WebKit source which is probably why it is not working as  
you'd expect.

If you're seeing what you consider to be buggy behaviour in Safari I  
would recommend filing a bug report with steps on how to reproduce the  
behaviour.  Based on the brief description of the issue you have  
provided it seems unlikely to be at all related to memory usage.

Kind regards,

Mark Rowe

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2413 bytes
Desc: not available
Url : http://lists.webkit.org/pipermail/webkit-dev/attachments/20080520/0ddb50a7/attachment.p7s