Xref: utzoo comp.os.os2.misc:258 comp.windows.ms:5683
Path: utzoo!attcan!uunet!cs.utexas.edu!yale!cs.yale.edu!spolsky-joel
From: spolsky-joel@cs.yale.edu (Joel Spolsky)
Newsgroups: comp.os.os2.misc,comp.windows.ms
Subject: Re: Presentation Manager and/or MS-Windows 2 or 3
Message-ID: <26360@cs.yale.edu>
Date: 26 Sep 90 23:59:35 GMT
References: <1990Sep26.204620.10987@odetics.com>
Sender: news@cs.yale.edu
Followup-To: comp.os.os2.misc
Distribution: na
Organization: Yale University Computer Science Dept., New Haven, CT 06520-2158
Lines: 60
Nntp-Posting-Host: giraffe.zoo.cs.yale.edu
Originator: spolsky@giraffe.CS.Yale.Edu

In article <1990Sep26.204620.10987@odetics.com> frank@odetics.UUCP (Frank Merrow) writes:
| I want to start a new project using a graphical interface at home using
| a "PC environment".  I have worked with Sunview before and so I am
| somewhat familiar with the concepts involved, but I have no direct
| experience with OS2 or MS Windows.  My understanding is that even though
| both OS2/PM and MS Windows were written by Microsoft, THEY ARE MILES APART
| so porting a program between the two is difficult at best.

Not entirely true. Porting a program between the two means replacing
all the Windows API calls with the corresponding OS/2 ones. The
trouble is that there is not a 1-1 correspondence, so this isn't
straightforward. Also, there are minor differences like OS/2 uses
bottom left as the origin and Windows uses top left. So porting is
indeed a headache but not impossible. The best approach is to write a
"layer" between Windows and your code; then you can isolate the part
of the project that you need to port. This is the approach most of the
industry uses for writing applications that run under, e.g., Windows
and Macintosh.

Microsoft is selling an alpha-release (read: literally HUNDREDS of
serious bugs) version of a "Software Migration Kit" which is supposed
to somewhat automate the conversion process. In reality the SMK bogs
down programs so they run much slower in OS/2 than in Windows. In
addition it is poorly supported. But, one day, the SMK might really
work. 

| Eventually, I would like my program to run under BOTH environments.  Because
| of this it seems appropriate to do the Windows version first.  First,
| Windows currently has a larger user base (and thus more customers) and
| second because in the Windows environment I won't get "suckered" into
| splitting my task using Multi-tasking concepts and so forth, because for
| the most part they just are not available under Windows to begin with.  I
| am assuming that the eventual port to OS2/PM would be easier than a
| port FROM OS2/PM to Windows (given that neither is likely to be "trivial").

Microsoft keeps telling people that OS/2 Version 2.0 will "run windows
applications unmodified". I am personally extremely skeptical about
this because (1) it will bloat OS/2 to about twice its present bloated
size (2) Microsoft, says InfoWorld, is pulling its programmers off of
OS/2 and putting them into Windows, and (3) Microsoft's current state
of disagreement with IBM over OS/2 is likely to dampen everyone's
enthusiasm for OS/2 for a long time. Especially if a future version of
Windows supports HPFS and threads.

| Once the decision is made to write the program in Windows, the next
| question is Windows 2.x or Windows 3.0.  The problem with Windows 3.0
| is that the customer base is (I presume) signifcantly smaller (unless
| MS has made the upgrade so cheap "everyone" is upgrading).  On the other
| hand if I write to 2.x, I am already "behind the times" and the
| product will age faster.

Wrong - the customer base for version 3.0 is MUCH bigger than the
customer base for 2.0, in fact, it is even bigger than the customer
base for Macintosh. Everyone threw 2.0 out the window; I doubt if you
could still get the 2.0 Software Development Kit any more. Besides the
fact that the technical benefits of developing for 3.0 are
significant.

Joel Spolsky
spolsky@cs.yale.edu                                     Silence = Death