Re: Algorithmic Complexity Attack on GLIB 2.2.1

[Sander Vesik <Sander Vesik Sun COM>]

On 31 May 2003, Owen Taylor wrote:

> I think the groundwork that would have to be done before we could
> look at what we wanted to change would be:
> 
>  - Instrument g_str_hash(), run a couple of large applications
>    (say, Nautilus, GTK+-2.0-based-gnumeric), collect histograms
>    of key size.
> 
>  - Write some medium sized benchmarks (say, construct a widget
>    heirarchy, display it on the screen, destroy it), try replacing
>    g_str_hash() to a universal hash and/or kernel-style Jenkins hash,
>    and see what the effect is on performance.
> 

One could use pbench here that allows you to measure gnome app startup
time as well as some important metrics like nautilus displaying the
contents of various sizes of directories. 

> Any volunteers?
> 
> Regards,
>                                                     Owen
> 

	Sander

OpenOffice.org - conquering the world 14000 PC-s at a time