TODO

There is much to do.

- Continue to clean up all the code.
- Make plmtty's interface self-explanatory.
- Add documentation.

There is an ugly hack in plmsend which stops plmcat then restarts it again 
in order to work around the plm's screwey flow control.  I'm not sure what 
the fix is.  At best, it would merely require a minor paradigm shift.  But 
at worst, I'm afraid it might involve creating a plm kernel module.  I don't 
plan to tackle this one immediately.

Currently, terminating x10toinst might leave an orphaned plmcat running 
that will not close automatically until data comes in on the PLM.  x10toinst
tries to kill these plmcats, but might miss some.

Assuming anyone is actually using plmtools and writing scripts, it might make
sense to have a forum of some sort for users to share scripts with eachother.

I am also kicking around the idea of splitting the C programs and the shell
scripts into two separate packages.  It might also make sense to have some
central repository of user contributed scripts.  If the files wind up getting
hosted on sourceforge, I believe contributed scripts will all have to use
sourceforge approved licenses, so bear that in mind.

Finally, the current interface to the C programs is through hex strings.  This
requires a lot of heavy lifting to be done in the shell scripts with sed, awk,
cut, etc.  I am considering the possibility of adding a higher level interface
to the PLM in the C programs.  This would reduce the burden on the shell
scripts and would translate into faster response times.  However, I think the
major source of delays is in the latency of the network, and not in the
overhead of the system calls.