Note that the following comments are about beta #2 -- I don't have the
final release yet, although it should be here any day now.

CS> The Visual Builder seemed a bit slow even on the Pentium 100MHz
CS> system I'm  using with 32MB RAM and a bus mastering SCSI adapter
CS> (NCR  825) with a Seagate ST12550N (Barracuda 2) hard drive.   Maybe
CS> 64MB RAM would help. And if IBM would ever get around  to releasing
CS> Warp SMP, I might add a second Pentium and see  if that would help.

 KS> Yep, this sounds like just the thing IBM should do to bring OS/2 into
 KS> the mainstream.  Build a development environment for C++ that requires a
 KS> supercollider to run. That ought to really encourage a lot of

Apparently Visual Builder is written in Smalltalk.  Perhaps this accounts
for its voracious memory appetite and slow performance. If that's the case,
it makes me wonder why anybody would want to use Smalltalk for developing
commercial software such as Visual Builder.

 KS> developers who can currently develop apps on any Windows platform,
 KS> including NT, with less powerful machines.

I don't mind spending an extra few thousand dollars on hardware to get good
performance with tools that make me more productive as long as the end
result (the programs other people will be using) are reasonably fast and
not memory pigs.  So far, I've been using ICLUI (renamed to OpenClass in
VisualAge C++) for user interface development and have been pleased with
both the development benefits and the end-result performance.

 KS> I know this isn't the OS Debate conference, but I'm having a hard time
 KS> understanding the logic behind VisualAge C++'s hardware requirements.
 KS> They're ridiculous.

Visual Builder seems like a giant memory pig compared to the C++ compiler
itself.  That's not to say it isn't useful.  Out of the two visual
programming tools I've tried for C++ (VisPro/C++ and Visual Builder),
Visual Builder is by far a more comprehensive and powerful tool.  I'm
hoping that the final release will be good enough to be something I can use
on a regular basis.

I guess what I'd like to see is a Smalltalk-like environment for C++
programming, changes to the C++ language to make it have Ada-style
separately compiled interface and implementation modules (rather than the
stupid *.h and *.hpp files and the preprocessor), and an incremental C++
compiler and linker that would allow C++ code to be quickly recompiled and
tested.

From the little I've heard of Taligent's programming tools due out on OS/2
later this year, they might come close to deliverying what I'd like to have
for a development environment.  On the other hand, I have no idea what sort
of CPU, RAM, and disk requirements they will have. The price, however, is
likely to be a killer -- it sounds like they may cost over $5000.  For a
single developer, that will make them very hard to afford.  But it also
sounds like that price allows as many developers as you want at a single
company to use the tools.  So if you've got 5 or 10 programmers and the
Taligent tools really are good,they would be very cost justifiable.  But I
guess we'll have to wait and see how this all turns out when they are
released for OS/2.


--- Maximus/2 2.02