REPORTING BUGS
==============

Please report any bugs encountered to

       Michael.Weber@cwi.nl


Please include enough information in a bug report that someone reading
it can reproduce the problem.


KNOWN ISSUES
============

[The gaps in the number sequence belong to old bug descriptions which
 have gone away.]

#1:
  The VM is relatively slow on Intel Xeon CPUs (compared to AMD64).
  Probable reasons: 
    * Cache effects; Xeon has only 512kB cache
    * Pipeline stalls; Xeon has a very deep pipeline, such that
      mispredicted branches are very expensive.  The inner loop of the
      VM does a lot of unpredictable branching...

#2: 2006-02-22
  Processes do not die in stack order.  No idea why this is needed at
  all.

  Also, if within the last step(sic!) a process-local variable is set
  and then the process dies, the never claim might not see it.  This
  is a more serious issue.  Ways around it: put "l: goto l;" at the
  end of such a process.

  http://spinroot.com/spin/Man/_pid.html:
  "Process instantiation numbers begin at zero for the first process
   created and count up for every new process added.
   [...]
   When a process terminates, it can only die and make its _pid number
   available for the creation of another process, if and when it has
   the highest _pid number in the system. This means that processes can
   only die in the reverse order of their creation (in stack order)."