On Wed, Dec 18, 2002 at 12:37:13PM +1100, matthew green wrote:
> 
> i think we could safely raise PID_MAX to 99999.  but beyond
> this and suddenly you've got pids > 5 numbers for the first
> time and i don't know what will break.  i've heard of a few
> horror stories from IRIX (they have huge pids).
> 
Probably not terribly relevant here because of the historical
background, but it has dawned on me recently (I am not a great
researcher, I may well have re-discovered something already
well-known) that the range 10001-99999 (for example) comprises 90%
of all available numbers less than 100000 while all its members
are exactly 5 digit long (no suppressable leading zeros).

Were I to re-invent Unix, I'd reserve the bottom 10% of the PID
range for internal purposes and only permit fixed length numbers
for normal operations.  Someone else can pick the upper limit.

Just a random thought :-)

++L