[Info-vax] MPA105: and RWAST'd

[Bob Koehler]

In article <774bd6e7-3c07-4cd0-8d51-8d60958965e9 at u28g2000prm.googlegroups.com>, Jojimbo <jjgessling at gmail.com> writes:
> 
> Thanks, Yes, I needed to empty out the pipeline so that the process
> can go away.  I changed to command to:
> 
> $   feob 563,grp104/match=3Dand | -
>     (read sys$pipe line ; define/job/nolog search_result &line ; -
>     copy sys$pipe nl: )  ! must throw away the rest of the lines
> 
> I am still puzzled by this however.  Searching around I found several
> examples where the emptying out the pipe isn't addressed.  Is it just
> because my data is long enough to get into rwmbx state?

   If you don't happen to fill up the mailbox before the child process
   exists, it won't be in RWMBX and can exit.  I'm sure there are lots
   of examples of this.

   It's quite straightforward to prove that correctly written programs
   always work correctly.  It's just about impossible to force
   incorrectly written programs to always fail.  If that could be done
   many security holes could be caught up front.