In a message dated 12-06-98, Mike Ruskai said to David Noon about "OS/2 C++
Programming?"

Hi Mike,

[snip]
 DN> I wouldn't say that I recommended it. I suppose it's alright if you
 DN> don't have anything better.

MR>EMX generates very tight and fast object code.  Faster than VACPP and 
MR>Watcom.

Not in my experience. Both IBM and Watcom produce faster code, even if it is
rather larger than that produced by EMX. [Does anybody still have Peter
Fitzsimmons's benchmarks from a few years back?] There is a thread in the
80XXX echo on this very topic (fast bloatware, under the title "NOP
accelerates code") at the moment.

MR>The downside to it is the difficulty in finding information, and
MR>that it has no IDE.

Very true. Most EMX users just use a make from the command line, though. I
use bake on the machine where I have EMX installed.
 
 DN> You will have to sign in as a "guest" but you can d/l
most stuff once
 DN> you have done that. It should allow you to d/l the Warp 4 toolkit, but
 DN> some of IBM's Web pages can be confusing. The last guy I told to do
 DN> this ended up with a d/l from Athens owned by IBM Greece. ... :-)))

MR>The toolkit is not available at the Guest level, nor even the Member
MR>level.  You must pay $300 for the Advanced level to have access to it.

Les Rhorer, formerly of Fido, was able to d/l the Warp 4 toolkit back in
September. He is the guy I referred to above. He used Guest level.

MR>IBM sometimes seems as if it's trying to prevent people from writing 
MR>software for OS/2, in a twisted way.

I think they would prefer people to pay for the privilege of pushing forward
the case for OS/2.
 
[snip]
 DN> Note that so far we have only discussed the C and C++ languages. There
 DN> is much, much more to learn about OS/2 programming, unless you are
 DN> content to limit yourself to command line interface (CLI) programs.

MR>The problem, however, is that there doesn't seem to be any affordable 
MR>languages other than C/C++.

ALGOL 68 and FORTRAN 95 compilers for OS/2 seem to be reasonably priced in
this country. Both suffer from the same lack of tools that blights EMX.

Have you priced the upgrade from VA C++ 3.0 to VA C++ 3.6? The 3.6 version
was not widely touted by IBM as the upgrade price was over US$3000. It was
strictly a corporate product.

As a PL/I programmer, I'm getting a kind of deja vu feeling about the
upgrade price from 3.0 to 4.0, when we finally see VA C++ 4.0 for OS/2.

MR>PL/I is now obscenely overpriced (I had planned on getting a copy to play
MR> with - no more, unless I win the lottery).

Did you try looking for Personal Edition at the site in the message I
forwarded here a few weeks ago?

MR>Smalltalk has always been overpriced.  COBOL is overpriced, and of
MR>dubious utility for micros.

SmallTalk is way too inefficient for anything worthwhile.

COBOL is too ugly for any platform.

MR>Unless, of course, you were merely commenting on the limitations of books
MR> that teach the languages :)

There are plenty of books on all sorts of languages. The problem is, none of
them focus on OS/2 implementations of the languages, and all the [remaining]
ones for the OS/2 API assume the reader is using C.

Perhaps I should take Disraeli's advice.

Regards

Dave

___
 * MR/2 2.25 #353 * Some mornings it's just not worth gnawing through the straps.

--- Maximus/2 3.01