So Is There Agreement On This, Now ?

Marcus G. Daniels marcus@cathcart.sysc.pdx.edu
12 May 1997 00:29:20 -0700


>>>>> "RJ" == Rainer Joswig <joswig@lavielle.com> writes:
>>>>>>> "CHG" =3D=3D C H Graham <cgraham@cgo.wave.ca> writes:

CHG> 2.  What about the suitability of "Plob" (or hacked version of
CHG> same) for persistence ?  (to be integrated down the road)
CHG> ... some of us may wish to look at this.

MD>  `Plob' uses a persistent store called Napier that costs =A3100.

RJ> Correction:

RJ> `Plob' uses a persistent store called Postore which comes (the
RJ> binary that is) for free with Plob.

>From the Plob user guide:

"POSTORE is the low-level persistent memory used by the 
persistent programming language Napier88."

My point is that we aren't getting freely redistributable source
code.  For this reason POSTORE is now useless; we must be able to fix
bugs.  Either the St. Andrews people need to be convinced to change
their licensing, POSTORE would need to be rewritten, or Plob adapted
to fit a new persistent heap like RVM.

Also realize that 1) there is a lot of indirection in the Plob system --
things go through TCP/IP RPCs, a Plob server, and then (the annoyingly
opaque) POSTORE and 2) Plob is definitely wedded to Common Lisp, and
less so to Harlequin LispWorks and Allegro.  

Finally, Plob is not suitable for a system-integrated transparent
checkpointing system, and I think some of us want that.