Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ucbvax!hplabs!hp-pcd!hplsla!jima
From: jima@hplsla.HP.COM (Jim Adcock)
Newsgroups: comp.lang.c++
Subject: Re: C++ 2.0 pricing (AT&T loses.)
Message-ID: <6590186@hplsla.HP.COM>
Date: 7 Jul 89 17:50:52 GMT
References: <2510@yunexus.UUCP>
Organization: HP Lake Stevens, WA
Lines: 66

>It occurs to me that the overall community (the side of which I sit by)
>will, in general, be better served by fewer versions of C++, each with
>support by a reputable vendor.

Define reputable vendor?  For me a reputable vendor is one which consistantly
provides me with high performance, highest quality products for competitive
prices year-in and year-out.  IE a vendor I can trust to do business with
over the long term.  A vendor who suddenly and unexpectedly radically changes
the terms of doing business does not inspire trust.

>The current situation, as I understand it, is that people get the sources
>of C++ to either port to their own platform (no help for them) or to fix
>bugs in the compiler and/or libraries.

Some of this is with people with more obscure platforms that aren't getting
formal C++ support yet -- and may "never."  While many people are on mainstream
machines, a reasonable fraction are not.

Also, a lot of this gets back to "trust" and ownership of compiler problems.
If one is spending $100Ks to $Ms on a development project, one cannot afford
to be stymied by compiler bugs, or unreliable vendors.  But presumably, big
commercial projects can afford to buy source at "any" inflated prices.

>I would argue that the cost (in person-power) of doing that level of
>source debugging and maintaining probably overwhelms the price that
>has been mentioned for source licenses (though in most of our organi-
>zations we find it easier to find payroll dollars than P.O. dollars).

I'm afraid, following up on this logic, that many managers will feel more
comfortable chugging along with low-priced [low-performance] C compilers,
rather than allowing their software developers the advantages of C++.
Scary prices are just too convenient an excuse to continue in one's set
ways.

>Fewer, vendor-supported, versions will actually increase the distribution
>of C++, plus give us better odds that random C++ programs will be
>portable across systems.

I don't believe this.  I believe the mega-multiple sources for C compilers has 
led to the explosion of C usage -- even though the vast majority of C compilers
are from relatively few vendors.  Likewise, I would *eventually* expect a 
similar pattern of C++ compilers.  A few mainstream suppliers, and lots of
little specialized nitch [sp?] players.

>I would say that, for better or worse,  C++ is probably going to be the
>"in" language of the 90's.  The price, performance, and robustness on
>Sun's, IBM PC's, and Mac's for binary distributions will certainly affect
>this.  So will the development of good user-interface building toolkits
>affect the impact C++ has on the world of programming.  However,
>the source cost won't.

These things can take longer than you think.  It may be that C++ becomes 
the language for the 2000's.  C++ is just marginally suitable for mainstream
programming today.  In a couple more years tools may become widely available.
And it can easily take 5 years for those tools to become widely used.

>To make C++ something other than a toy of the priveleged - that seems to
>me to be a goal worth shooting for.  It is interesting, is it not, that
>that seems to be all of our goals, and yet the method is so controversial?

I think one area that needs to be re-thunk is C++ usage in universities.
Unix became widely distributed though the support of universities, many
computer vendors heavily discount or donate their machines in order to
secure a university presence, etc.  I'd like to see an approach taken that
ensures a lively presence of C++ in universities.  I think g++ is useful in
that regard, but not sufficient.