Re: Ada Compiler for Educational Use

[ecragg@GMUVAX.GMU.EDU ("EDWARD CRAGG")]

> Date: Thu, 12 May 88 09:44:03 PDT
> From: pattis@june.cs.washington.edu (Richard Pattis)
> Subject: Ada Compiler for Educational Use
> 
>   I have been teaching the CS-1 and CS-2 courses here at the University of
> Washington in Modula-2 for the past 4 years.  I am now strongly considering
> teaching Ada in these classes.  I would like to solicit recommendations for
> compilers: we use a large VAX/VMS for instruction, but will also ultimately
> need compilers for our own department's machines (mostly VAX/ULTRIX and Suns).
> I am looking for a system with short compile times and good error diagnostics;
> fast run times are not as important.  We typically have 20-40 students logged
> on at once.  Also, I am interested in other language related tools, such as
> language sensitive editors, debuggers, and profilers. I hope to direct the
> Computer Center while it examines a few compilers early in the summer, and
> then choose one for use this Fall.  I'd appreciate any relevant comments.

  I have used DEC Ada, TeleSoft Ada, VERDIX Ada, and the ALS/N CAB all under
VAX/VMS.

  If you have the funds, DEC Ada gives by far the most detailed diagnostic
messages and has the best librarian.  Furthurmore, its documentation is superb,
if expensive (LRM @ $50, Developing programs guide @ $25, and Runtime guide @
$25).  Finally it is fully integrated with both the DEC Debugger and Performance
Coverage Analyzer.   I have found that class room demonstrations using the
debugger have been most effective both in showing execution and showing the
effects of optimization.  It would be my system of choice. 

   TeleSoft Ada did not do anything.  I ended up canceling the class 1/2 way
through the course because it was so poor.  The TeleSoft reps were not very
helpful either which bothered me because were were a cash customer -- paying
$15K as I recall.   I recommended that we attempt to get a refund, but the
powers that be chose not to. 

   Verdix was so vanilla that it left no real impression.  We tested a VADS
for several weeks, and got programs running but we did not pursue the system
furthur.  As best I remember the documentation was not the best and it took
quite a few false starts to get the programs running.  Since we were running
test programs which had been developed under other systems, we did not have
many diagnostics.

   The ALS/N CAB is the system we are using, for funding reasons.  (Its hard to
beat $75). My biggest criticism is the quality of the diagnosic messages and
the interpretation of how TEXT_IO GETs and PUTs are interlaced.  (They are
interlaced at the line level, rather than at the character level). This latter
problem we are resolving by writing an IMMEDIATE_IO package which does what its
name implies by calls to QIO.  The July release is advertised to include an
interactive debugger. 

-------------------------------------------------------------------------------
Edward E Cragg                Bitnet:   ECRAGG@GMUVAX
                              Internet: ECRAGG@GMUVAX.GMU.EDU