home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
OS/2 Shareware BBS: 14 Text
/
14-Text.zip
/
ZDWNOS.ZIP
/
ZDWNOS.TXT
Wrap
Text File
|
1992-11-02
|
16KB
|
251 lines
PC Magazine
One Park Avenue
New York, New York 10016
Re: Comparison of Windows 3.1 and OS/2
November 10, 1992, Vol. 11 No. 19
Dear Editor:
Having just read your comprehensive comparison of Windows 3.1 and OS/2, I feel a
few comments and corrections are appropriate. My comments are based on six
months of using OS/2 on a Northgate 486/25, with 16MB RAM, an ATI Graphics
Ultra card, an Adaptec 1542B adapter controlling two SCSI drives, a SoundBlaster,
an Epson 300-C scanner with a proprietary controller, an NEC CD-ROM drive attached
to a Trantor controller, a PPI 2400 baud modem, and a Mountain TD8000 tape drive
with a proprietary controller. This is hardly a vanilla clone, let alone an IBM machine.
All of the peripherals run properly, with the exception of the tape drive and CD-ROM.
Mountain is wavering with respect to promised OS/2 software for the tape drive; it
is my understanding that they already produce OS/2 software for other tape drives
they manufacture. In the case of the CD-ROM, I cannot get sound, but I don't use
it for that purpose, anyway; all other CD-ROM functions perform optimally.
The Introduction
In the first paragraph of Sheryl Canter's introduction, she asks the question, "Should
the choice be made strictly on technical merits, or should each company's marketing
track record also be considered". Stated that way, the answer is obviously that both
should be considered. But in the discussion of that issue, only Microsoft's marketing
success with respect to Windows is considered; there is no discussion of Microsoft's
abandonment of OS/2 - and OS/2 customers - when Windows offered Microsoft a
superior marketing opportunity. While no one wishes to back a loser and thereby
become orphaned, no one wants to back an operating system that may be changed
at any time, effectively making the user's installed base of applications obsolete. If
marketing skill is important, as it clearly is, then so is vendor reliability. The article
emphasizes one issue, while ignoring the other.
Later in the introduction, Ms. Canter explicitly points out that OS/2 1.x failed because
it was not backward compatible. Yet there is no comparable statement that Win 1.x
and 2.x failed for the same reason, or that Windows NT will face the same problem.
On page 207, Ms. Canter commits her most egregious error, stating that, "video
resolution under Windows is currently limited to VGA". This is patently false.
Windows is available under OS/2 at any resolution supported by OS/2. For example,
I am currently running OS/2, including Windows sessions, at 1024x768x256 without
significant problem. The restriction that exists is that on most video adapters,
Windows under OS/2 must be run full screen in high resolution video modes, just as
it must in high (and low) resolution under DOS. Windows applications can be still be
run in windowed mode within the Windows-OS/2 session, but the Windows session
as a whole occupies the entire screen. The only difference between high resolution
Windows-OS/2 and Windows-DOS is that under Windows-OS2, the user can switch
out of Windows to do non-Windows tasks, without terminating Windows programs
that may be running; under Windows-DOS, Windows must be shut down completely.
This hardly seems an advantage for Windows 3.1.
In addition, if a user is willing to operate under VGA, Windows applications can be run
directly in an OS/2 window on the OS/2 desktop ("seamless Windows"). This ability
is being extended beyond VGA as high resolution drivers allowing seamless Windows
become available. Several such drivers will be included in the Service Pack for OS/2.0
being released by IBM within the next few days. Other seamless drivers are in
development, both by IBM and other vendors (some with more ambition than others,
admittedly).
Next, Ms. Canter remarks that Windows "has a tendency to crash". Unfortunately,
that is the extent of the discussion. I am not a Windows 3.1 user; I did use Windows
3.0 regularly for 18 months. I can confirm that Windows "has a tendency to crash".
But I would like to know more about this "tendency" in 3.1; according to Microsoft,
3.1 is supposed to be completely stable. Why wasn't the matter given further
discussion? There was certainly substantial discussion of the supposed "instability"
of OS/2.
Finally, the introduction states that, "Installation and setup require a high level of
patience and technical competence. Worse, OS/2's numerous quirks are bound to bite
you if you try anything unexpected." As to installation and setup, as noted above,
I do not have a vanilla setup, but I had no problems whatsoever with installation, once
I read the instructions. I just inserted the disks, one after another after another after
another ... . It was boring, but hardly difficult. I did have to call Trantor's BBS to get
CD-ROM drivers. Contrast this with my Windows and DOS experience. Upgrading
to DOS 5.0 took three full days on my old, 286 based system, with one problem after
another arising. Win 3.1 won't install properly at all on my current system: my
swapfile always ends up corrupted.
As to "numerous quirks", if that means that you do things under OS/2 differently than
you do under Windows, that is true. It must have taken me at least 5 or 6 days to
become comfortable with OS/2. If the phrase refers to something more ominous, I
have to say I have not encountered it. There were certain specific, repeatable bugs
which I encountered, all of which involved the use of specific applications. All of
those which interfered with my use of the computer have been fixed by IBM; I would
have appreciated more speed, but they were much faster with the fixes than
Microsoft was about fixing Windows 3.0's much more significant problems.
Certainly, OS/2 has not presented me with the number of inexplicable problems I
encountered under Windows 3.0, where the machine - and my work - would suddenly
stop for no apparent reason. I finally reached the point with Windows (as did many
others), where, if it was important to finish without interruption, I exited to DOS first.
The Performance Comparison
The Performance Tests were seriously flawed. First, why did you compare different
version levels of Windows and OS/2 applications? This was supposed to be a
comparison of the Operating Environments, not the applications. A proper test would
have been to compare similar version levels of the programs, such as Microsoft Word
3.0 for both platforms. Given the improvements which have occurred in the
applications, the results you obtained are meaningless as a measure of Operating
Environment performance.
Second, I'm not surprised at your test results when single tasking WordPerfect 5.1
for DOS; my experience is not the same, but I have a faster machine, with more CPU
cycles to spare. OS/2, to accomplish its goals, must include more overhead than
Windows. For optimizing performance of a single DOS programs, the DOS command
line is the correct environment.
Finally, why did you elect to use Windows programs for the multitasking test?
Windows 3.1 is optimized to run Windows applications, while OS/2 is optimized to
multitask everything available for Intel chips; of course Windows was faster at
multitasking Windows programs. Why didn't you use DOS programs (where OS/2
would win, I suspect, hands down), or a mix of OS/2, Windows and DOS programs?
How many people who are interested in multitasking will use only Windows
applications? This test is worthless as a measure of "real world scenarios".
The Windows Review
Mr. Bonner's article is certainly forcefully directed: 21 paragraphs, of which 12
contain variations on the same theme. "There is more third party support for
Windows", he tells us. And tells us. And tells us ... . Don't you think that five or six
times would have sufficed? It is not news that OS/2 has been out for only seven
months, and that Windows 3.x has been available much longer; how much third party
support did Windows 3.0 have after seven months (Microsoft's own headstart
applications don't count; utilities that are not needed with OS/2 don't count)? That
information would give us some indication of OS/2's chance of success in the market
place.
In the review, Mr. Bonner also states that "Given its stable, speedy performance ...,
Windows 3.1 is the obvious choice for a desktop environment". Did Mr. Bonner and
Ms. Canter talk to each other during the preparation of these articles? Given their
conflicting opinions on stability, don't you think that more discussion of the subject
is necessary?
The point is made in the review that Windows is superior to OS/2, in that if a problem
arises with Windows, one can boot without loading Windows and fix the problem
from DOS. I have three remarks. First, one can't do that if Windows is loaded by the
autoexec.bat file; in that case, one must boot from floppy, just like OS/2. Second,
if Windows fails under OS/2, one can exit Windows and fix the problem from the
OS/2 Desktop. Third, the importance of this "feature" depends on the frequency of
problems, doesn't it?
Mr. Bonner attributes the speed advantage shown by Windows not to your selection
of tests and test applications, nor to the admittedly heavy overhead a preemptive
multitasking environment necessarily entails, but to the fact that "much of OS/2
consists of 16-bit code, while Windows 3.1 makes use of generous helpings of 32-bit
code ...". This is either incredibly bad writing, or an attempt to make readers believe
that Windows 3.1 contains more 32-bit code than OS/2. Such is not the case; the
exact proportion of 32-bit code in the OS/2 kernel is IBM proprietary information, but
it has been made clear by IBM programmers that a large portion of the remaining
16-bit code is in the Graphics Engine (a full 32-bit version of which will be available
within a week), the High Performance File System, the various device drivers shipped
with OS/2, and (of necessity) the Windows and DOS subsystems.
Finally, Mr. Bonner states that Windows NT will solve Windows "major weakness".
How many applications will be available for OS/2 and NT when NT ships? How much
hardware will NT require to obtain performance equal to that of OS/2? How stable
will NT be? How many DOS and Windows applications now produced by users
themselves and by companies other than Microsoft will run under NT? Without
concrete answers to these questions, how can NT be considered a solution to
anything?
The OS/2 Review
The review of OS/2 by Lenny Bailes is, for the most part, competently written, if
somewhat less enthusiastically and aggressively than Mr. Bonner's review of
Windows. But I do have a few comments.
Overall, I feel that the article places far too much emphasis on installation problems,
since I did not encounter them. I must admit, however, that they do occur, although
seemingly more often to journalists than to the rest of us.
The first page of the review has a picture of an OS/2 Desktop screen. The caption
reads, "Some DOS apps, like SuperStor, cannot run in OS/2's bundled DOS
environment. They require a standalone version of DOS as a separate session." This
statement is certainly true, but what does it have to do with the image to which it
refers? Why is it sitting out there in bold face type? And perhaps your readers would
be interested in knowing why programs such as SuperStor and Stacker can't be
allowed to run in a memory protected environment.
The second page of the article is garbled, presumably because of a last minute
decision that the appropriate place for an NT sidebar was in the middle of the OS/2
review (why weren't future plans for OS/2, such as Win 3.1 support, portable OS/2,
the 32-bit GRE and OS/2-AIX given a sidebar in the Windows review?). However, as
best I can decipher the page, it appears that it was necessary to replace the I/O card
in your clone system with a newer, post-1990 card providing for hardware interrupts,
a feature required for true (if heretofore unused) PC compatibility. When was the
original card made, and did it omit hardware interrupt features? If so, why is this an
OS/2 problem? If one buys cheap hardware that doesn't meet specifications, one has
to live with that decision.
Mr. Bailes also states that AutoCad (Rel. 12) will not run under OS/2. This has been
corrected through the combined efforts of Autodesk and IBM, according to the most
recent postings on CompuServe.
The same error made by Ms. Canter concerning high resolution Windows is repeated
by Mr. Bailes. I refer you to my previous comments.
The Summary of Features
The "Price" listing for Windows should add the cost of DOS; since you use the full list
price for OS/2, which is only paid for entirely new installations, the same basis should
apply for Windows. It cannot run without DOS, which I believe has a MSRP of at
least $99.00. Your hard disk space listing for Windows also omits the requirements
of DOS, I believe. Since I don't run Windows any longer, I can't be sure.
The Editor's Choice Award
I won't argue with your selection, since it really is a matter of personal preference.
I do argue with the validity of your stated reasons.
Your most important reason is the large number of Windows applications which
currently exist; this is certainly consistent with the review of Windows. But you
discount the fact that OS/2 runs these applications because "in doing so you do not
take advantage of OS/2's unique capabilities". Why is that an advantage for
Windows? Windows 3.1 doesn't provide those capabilities, either. Further, following
that reasoning, we should not count as Windows 3.1 applications those programs
which were originally written for Windows 3.0 and which have not been upgraded to
use the 3.1 feature set. If that's the case, then there are as many native OS/2
applications as there are Windows 3.1 applications (once again excluding the "head
start" applications produced by Microsoft).
Another of your reasons for selecting Windows 3.1 is that it "delivers faster
performance ... when running today's 16-bit applications." This statement is true for
the first application, due to OS/2's protection overhead, and for pure Windows
multitasking, but you did not test pure DOS and DOS + Windows multitasking, which
is the scenario for which most people would buy OS/2. How many users own
Windows applications and nothing else? Would Windows even come close in such
test?
You also justify your choice of Windows 3.1 by stating, "it even does well with
mission-critical tasks." You say this in the face of Ms. Canter's observation that
Windows "has a tendency to crash". What exactly do you consider a mission critical
application? And in the absence of serious stability comparisons, how can stability
be said to favor Windows 3.1, given the widespread complaints among its users
(including John Dvorak and now, rumor tells me, Bill Machrone)?
Further, you state that installation problems on non-IBM hardware and "quirks ...
[that] can cause the system to lock up unexpectedly" prevent you from recommending
OS/2. I can only say that our experiences differ dramatically. I have had many fewer
problems with OS/2 than I had with Win 3.0.
Finally, you pick Windows 3.1 because OS/2 support is uneven. Do you seriously
contend that Windows support for users is better?