Re: QNX subprocess bug

[bob]

I have observed PIDs very high in the 32-bit space (most user processes) and 
quite low (most system processes) and very little in between.  Right now I 
don't 
have access to my QNX system so I can't give you any real examples.  QNX is not 
open source so there is no obvious way to tell how the PIDs get generated.

The float (I take it we're talking "double" here) sounds like the best 
solution. 
I'll hack it into the code and see what happens.

Thanks!

>Date: Thu, 30 Mar 2006 22:11:02 -0500
>From: Richard Stallman <address@hidden>
>Subject: Re: QNX subprocess bug
>To: address@hidden (Kim F. Storm)
>Cc: address@hidden, address@hidden
>X-BadHeader: plain; charset=ISO-8859-15
>Original-recipient: rfc822;address@hidden
>
>    I don't know if it is true, but someone recently told me that even the
>    Linux kernel will use 32-bit pids if you happen to create more than
>    32000 processes at the same time.
>
>Does that really mean all 32 bits get used?
>
>There are not a gigantic number of places in the code that call
>process-id.  But at least one prints the pid in decimal, so using a
>cons cell would not work well.
>
>Using a float would work.

                 | 
Bob Bramwell     | Many people would sooner die than think; 
ProntoLogical    | in fact, they do so.
+1 403/861-8827  |  - Bertrand Russell
                 |