At 03:31 PM 13/4/95 METDST, Chris Breemer wrote:
>        "Why is there no C++ compiler planned/available on MPE/iX ?"
 
I would qualify that to "Why is *HP* not making a C++ compiler available on
MPE/iX? ".
 
Dr. Harry Sterling, speaking at IPROF, addressed this matter in both his
HP3000 Future Directions talk, and later at the Management RoundTable.  I
apologize that I don't have verbatim quotes, but two of his statements he
made seemed, to me at least, to be contrary to each other.
 
First, he said that HP was actively working making the HP3000 more "open".
For HP's part this involved providing the tools necessary to aid 3rd parties
in migrating their software to the HP3000.  It also required supporting the
popular standards adopted by their customers and prospective customers.
HP's records is very good in this regard and there are numerous examples,
especially in the last year, to attest to this.
 
>At Uniface we port to many platforms. MPE is the only one on which no C++
>compiler is available, or even planned. Yes, even on IBM MVS there is one !
>This is actually beginning to hold up our natural migration path from C to
>C++.
>
>If HP is totally committed to open up the HP3000 for (UNIX) apps they should
>provide a C++ environment. If not the HP3000 will miss out on a certain
>number of object-oriented applications and this problem may well aggravate in
>the next few years to come.
 
This is the proof, if it wasn't intuitively obvious, that one if not many
major third parties *are* using C++, and that it is a standard for current
and future development by many vendors.  C++ is the principal choice for
many sites if only as a better 'C'.  But Dr. Sterling's considerations
failed to mention existing 3000-based vendors or installed customers.
 
I don't know how many vendors are using 'C' versus 'C++'.  Most C++
compilers I've seen seem to be able to support 'C' pretty well, but the
reverse is not true.
 
Dr Sterling's second statement was that HP were definitely *not* working on
a C++ compiler, had no intention of doing so.  HP is satisfied that a public
domain GNU C++ compiler was being worked on by third parties.
 
So how many vendors are going to put their business life in the hands of a
public domain, unsupported compiler?  Ignoring concerns on platform
efficiency, the market for *critical* business applications would be more
than a little suspect of programs written with shareware.  "Will this work
on the next 5.0 PushMe-PullYou Release?", they ask.  "Who knows!".
 
 
At 01:57 PM 13/4/95 -0400, Scott Herman wrote:
>What C++ pseudo standard would you suggest this compiler adhere to?
>
>Unfortunately, there is not yet an ANSI (or any other) standard for C++
>(sigh)
 
The ANSI C standard is extremely stable.  In fact, there haven't been any
changes since it was adopted!  On might say, "As stable as a tombstone.".
 
The ANSI C++ standardization committee, on the otherhand has been much more
active, listening to its members and the public in general.  The "C++
Annotated Reference Manual" by Ellis & Stroustrup, is the accepted ANSI Base
Document, with Revision 3.0 the most stable.  C++ is a living and growing
language: this vitality is an asset, not a liability.  Granted, this does
make it more difficult for the compiler developers, but not impossible.  HP
bear witness to this with their C++ compilers for HP-UX and their
committment to widely regarded SoftBench tools.
 
At 11:42 AM 13/4/95 -0400, Denys Beauchemin wrote:
>HP keeps telling us they need a business case to do anything on the HP3000.
> As this does not seem to be a problem to do things on the HP-UX platform, I
>sometimes question HP's true feelings with regards to the longevity of the
>HP3000.  Then I look at what they have done on IMAGE, ALLBASE, POSIX and I
>find myself being ambivalent.  Why can't this enthusiasm be found across the
>board when it comes to languages.
 
Sometimes I too feel HP is dropping the MPE/3000 ball.  It is all well and
good to explain there is no market demand there for something.  This is
hardly surprising for something that is not currently there, and is being
denied a potential future by HP Management.
 
I guess for many of us we are struck by HP's oft repeated statements that
they wish to make the 3000 truly Posix compliant, in order to encourage
vendors to migrate.  Why then can't make the committment to even migrate
their own compilers?  I can hear the prospective vendors saying, "If its so
damn good, why don't you do it yourselves?".
 
>I am sure that HP is paying attention, we just have
>not reached critical mass yet.  (My guess is we are still a long way from
>this point but I am in for the long run.)
 
I, for one, am being stymied by the inability to migrate C++ code from other
platforms to the HP3000.  If there were C++ compiler, I would order it on my
next 3000.  I'm not even a big fan of 'C', really.  But in any case, there
is no C++, there is no work being done on late-binding class libraries for
the 3000, and most HP3000 Cobol shops will not be switching in anycase.
 
I feel bad that I didn't take the opportunity to address this personally
with Dr. Sterling at the IPROF, for he made himself available many times.
Perhaps he still is listening.
----
Jim "seMPEr" Wowchuk           Internet:    [log in to unmask]
Vanguard Computer Services     Compu$erve:  100036,106
 _--_|\                        Post:        PO Box 18, North Ryde, NSW 2113
/      \                       Phone:       +61 (2) 888-9688
\.--.__/ <---Sydney NSW        Fax:         +61 (2) 888-3056
      v      Australia