Re: [PATCH] O12.2int for interactivity

[Timothy Miller <miller@techsource.com>]

Rob Landley wrote:
> On Tuesday 05 August 2003 18:49, Timothy Miller wrote:
> 
> 
>>Or closer to the point:
>>
>>"For each record player, there is a record which it cannot play."
>>For more detail, please read this dialog:
>>http://www.geocities.com/g0del_escher_bach/dialogue4.html
> 
> ...
> 
>>The interactivity detection algorithm will always be inherently
>>imperfect.  Given that it is not psychic, it cannot tell in advance
>>whether or not a given process is supposed to be interactive, so it must
>>GUESS based on its behavior.
> 
> 
> Another way of looking at it is that every time you remove a bottleneck, the 
> next most serious problem becomes the new bottleneck.
> 
> Does this mean it's a bad idea to stop trying to identify the next bottleneck?  
> (Whether or not you then choose to deal with it is another question...)


No.  It just means that it's possible to produce artificial loads that 
break things, and since those artificial loads won't be encountered in 
typical usage, they should not be optimized for.


Mind you, we prefer that the worst case "record you cannot play" doesn't 
have TOO much impact, because we don't want people writing DoS programs 
which exploit those artificial cases.