Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site aplvax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!harpo!seismo!rlgvax!cvl!umcp-cs!aplvax!mes
From: mes@aplvax.UUCP
Newsgroups: net.lang.ada
Subject: Summary (incomplete) of Ada Compilers
Message-ID: <518@aplvax.UUCP>
Date: Wed, 22-Feb-84 08:36:13 EST
Article-I.D.: aplvax.518
Posted: Wed Feb 22 08:36:13 1984
Date-Received: Thu, 23-Feb-84 04:43:19 EST
Organization: JHU/Applied Physics Lab, Laurel, MD
Lines: 38


I'm not an expert on what's currently on the market, but I'll tell you
what I know (or think I know).  Rumors about future products abound
but the ones that are actually living somewhere in the bowls of a computer
are as follows:
    Telesoft Ada - Runs under Unix, generates (or will soon) code for
                   VAX, 68000.  Current version is a very reasonable
                   subset ( but no task types, or generics).  Rumor is
                   that a new version is about to be validated.

    York Ada -     Runs under Unix on a VAX.  Seems to be rather fast 
                   compared to many others.  My info. about this one
                   is limited.

    NYU Ada-Ed -   A horrendously slow but validated compiler.  It's 
                   not really that bad a tool if all you want to do is
                   do syntax and semantic checking on programs, bbut if you
                   want to actually run them,  better bring a good book.

    Intel iAPX-432 - This compiler is fairly fast comparatively, but eh
                   development environment is quite cumbersome, and requires
                   an iAPX-432.

    Western Digital - This compiler is validated, but from what I understand
                      is incapable of executing a program larger than a 
                      few hundred lines.

    ROLM Ada   -   This is the only validated compiler that seems anywhere
                   near ready for serious use (if validation is required),
                   but still has a few practical bugs in it (e.g. there is 
                   a limit of 32 on the number of tasks that can be built
                   into one program.

                         
                                 Happy Hunting,
                                 Mark Schmid
                                 Johns Hopkins Applied Physics Lab