There is only one tech report available via e-mail from 1989.

TR37-1989.ps	The ALPS Kernel  by Prasad R. Vishnubhotla et al

   ALPS is a parallel programming environment with emphasis on the 
portability of application programs across parallel machines while ensuring 
their efficient execution.  The programmer need not be concerned whether the
machine is a shared memory multiprocessor or a processor network, or about
the interconnection topology of the machine.  To achieve both portability
and efficiency, we use a new programming paradigm.  Central to the ALPS
paradigm are two concepts:  multiplexer channels and managers.  A multiplexer
channel is a many-to-many communication medium shared by a group of processes.
When a consumer wishes to receive a data item on a channel, it can non-
deterministically be given any one of the values on the channel.  Multiplexer
channels are typically used as carriers of work-to-be-done, and the implemen-
tation can exploit the data nondeterminism to provide automatic work-load
balancing by dynamically partitioning the messages on the channel among the
consumers.  ALPS supports object-oriented programming wherein objects can be
shared by concurrent processes.  The manager mechanism provides concurrency 
within a shared object.  If there is a manager process for an object, each
call to an object's procedure is intercepted by the manager.  The manager 
can decide which of these calls should be allowed to progress concurrently
and which of them should be queued up until a future time, thus implementing 
any necessary synchronization.
   This document describes the ALPS kernel which provides a small set of 
primitives that support the ALPS paradigm and are callable from C programs.
Initial versions of the kernel have been implemented on an Encore Multimax, 
a BBN Butterfly, a Transputer network, an Intel iPSC and a network of Sun 
workstations.

Key Phrases:  Parallel Processing, Programming Paradigms, Operating Systems,
	      Programming Environments, Machine Independence, Cooperating
	      Processes, Object-Oriented Programming.