Newsgroups: comp.lang.smalltalk
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!in1.uu.net!ncrgw2.ncr.com!ncrhub2!attatl!alewis
From: adam.lewis@atlantaga.ncr.com (Adam Lewis)
Subject: Re: The Future of Smalltalk
Sender: news@attatl.AtlantaGA.NCR.COM (nntp network news service)
Message-ID: <Dn8vzL.Mzr@attatl.AtlantaGA.NCR.COM>
Date: Fri, 23 Feb 1996 20:10:57 GMT
References: <reed-0702961024210001@stic.pdial.interpath.net> <4fueip$7rj$2@mhade.production.compuserve.com> <4ggq2e$mca@news2.ios.com>
Nntp-Posting-Host: alewis.atlantaga.ncr.com
Organization: NCR Corporation Retail Systems Group
X-Newsreader: News Xpress 2.0 Beta #0
Lines: 57

In article <4ggq2e$mca@news2.ios.com>, vlad@gramercy.ios.com (Vlastimil 
Adamovsky) wrote:
>Joe Hughes <74671.617@CompuServe.COM> wrote:
>
>>I'm too busy to do the whole survey but I have 5 years of ST with 
>>ParcPlace, ENFIN and Digitalk.  I feel the following:
>
>>The future of Smalltalk heavily depends on growing the supply of 
>>Smalltalkers.  The only way to do this is for the IBM and PPD to 
>>release some kind of beta or free Smalltalk on the Web.  I know 
>>that GNU is 'free' but that's not what the corporations use.
>
>>I am very upbeat about Smalltalk, we are selling VisualWave 
>>Weblications as fast as we can make 'em.
>
>>For all the bluster with C++, my own company has thrown it out in 
>>favor of Smalltalk, Delphi and Notes.  Yes, C++ has all the 
>>features, but it takes too long. 30 persons shops which drive the 
>>industry don't have the time to build their own libraries.  MFC 
>>is crud-run NuMega's bounds checker on it and count the leaks.  
>>We hired the head of Lotus R&D-did you know that they write ALL 
>>their C++ classes from scratch since MFC is so bad.
>
>MFC can be really bad. Does it mean that C++ is bad?
>If Smalltalk is good, does it mean that you can easily find the class
>library you need?
>
>What kind of Smalltalk do you use? 
>
>All those language are very incompatible, which I consider as very
>wrong. Furthermore, GUI part of VisualWorks is outdated and primitive.
>GUI part of Digitalk Smalltalk is excellent. 
>
>So, the future of Smalltalk depends on the feedback we (the
>programmers) will get to the Smalltalk vendors. Being satisfied with
>bad incomplete things can only cause harm to those languages.

Two points in response:
1.  I think the answer to your query about bad points of C++ can be extracted 
from Joe's original message.   If anything, dealing with the subtle nuances of 
how C++ allocates objects in memory is a big minus for the language.  Joe's 
comment about running NuMega BoundsChecker against MFC is apt.  It's scary to 
look at the results you get when you do that.    Another example.  The last 
C++ project I worked on was contained about 3 KLOC.   Back tracking back 
through the results of testing, we found that we averaged about 1 memory 
related fault (leaks, dealing strange nuances when casting, etc.) about every 
1KLOC.    It helps to either have a language be real picky about these types 
of things (languages like Effiel) or have a language to tries to deal with the 
problem down in your runtime (Smalltalk).

2.   One of my personal hopes is that ParcPlace - Digitalk will adopt more the 
VS philopsophy in their GUI framework.   Everything I've heard tends to 
indicate that PP-D is thinking in these regard for the "Grand Convergence".  
Let's take a wait and see.

AWL

