Thanks for the b0rk3d code!

[Fairlight]

I'd just like to thank fP-Tech (and the incarnations before them) for the
utterly horrible error-handling code, whereby STDIN is polled as fast as
can be for [Enter], even though it will never come in a cron, CGI, or other
headless context.

I've always contended that this should -at least- use select(), rather than
be a polling loop.  Ideally, it should check to see if STDIN is a bad file
descriptor, and simply exit if it is.

However, this egregiously sloppy coding provided a useful solution for a
different issue.  I needed to artificially inflate the load average to test
software which will notify me when filePro does exactly this (and other
load-hijacking circumstances).  In knowing that fP will jack up the load
average this way, I was able to use the technique to write a load average
stress testing harness which will gracefully ramp the load to a pre-defined
level, without running away with the spoon.  I actually needed to emulate
this in a controlled fashion, and the technique works a treat, when used
properly and intentionally.

So, thanks...I think.  This is probably the first good thing to ever come
from the debacle that is filePro's polling loop on STDIN.

m->
-- 
Audio panton, cogito singularis.