Re: RBTree implementation and thread structure

On Mon, Apr 14, 2008 at 2:50 PM, Rene Gollent <[email protected]> wrote:
>  Right, I'm not disputing that :) I'm just saying that I don't see any
>  meaningful way of evaluating "scheduler X results in better
>  responsiveness than scheduler Y" programmatically :)

Heh, and I wasn't saying that to you specifically, but used that hook
to shamelessly plug my view on the matter. >:D

Can't we use 'hey' or something similar to send UI events to
applications? AFAICT those applications must explicitly set scripting
hooks, but perhaps there's something else that can interact with UI
controls that I'm not aware of. Else the Big Fish apps like Mozilla
(Big Reptile?) can't be tested mechanically. Then yes, a double-blind
test with some significant sampling of machine configurations is the
way to go. Thank God for the Internet ;) as it should be only a matter
of compiling different kernels and shipping them to the testers,
probably tampering with some random data to generate different MD5s
(change linking order, perhaps? Does GNU ld even allow this or will it
internally reorganise the objects so they always land in the same
order on the final image?), and then we keep a private database of
kernel signatures.

Uh, I guess I'll leave those details to the website backend wizards...


Cheers,
A.

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone