🚀 go-pugleaf

I've debated for sometime with myself whether I should write this message, but
as yesterday I finally discover that the >$700 development tool set we
bought from IBM -- DOES NOT HAVE A PROFILER, I decided I really did need to
write this.  (Yes, I know they now sell the tools for $295, but then
that's about what they're worth -- and we paid >$700 -- guess that's
what we get for being anxious to develop for OS/2.)
I ask that somebody from IBM please post this on you internal bulletin board.
And I really do expect a response, preferably from someone fairly high up,
and preferably addressed to developers in general, not just me.

This posting reflects my personal opinion, and not necessarily that of the
University of Utah, or the spin-off company associated with the research
group I work for.

The following is a description of my adventures in trying to develop
a product for OS/2 2.0, and more recently for Windows NT.

Back in late January, I started following closely comp.os.os2*.  It seemed
like the "impossible" was about to happen.  OS/2, which always looked
good on paper, was about to have a good implementation.  Further all that
286 nonsense had been dropped.  I had only just gotten back into the PC
world a few months before as I saw in Windows 3.X that finally PC's were
beginning to get "usable" (read capability near the level of UNIX), and seeing
as they already had neat stuff of their own (good word processors, cheap but
still real database support, great games, full-featured GUI's [X is not
full-featured]), that the capabilities
of Windows combined with the power of the ever cheaper 386/486's meant
the world was about to change, and that UNIX may have missed its chance.

Our research group is mainly UNIX based, and so I got quite a bit of
fun-style "abuse", in doing this... and when I announced that OS/2 2.0
was "almost as good as UNIX, better in some ways" the response was somewhat
more than skeptical.  Still things looked good enough that we decided to
postpone some work I had scheduled for Windows 3.X, and push on to OS/2 2.0,
as that was clearly a better platform for our product.
So far so good.  Finally a non-toy operating system in the PC world.

In early February I started trying to contact IBM.  For whatever reason
the people in Boca were not answering the phone.  A month later I was finally
able to start development, using EMX and a copy of the LA release a local IBM
guy gave me -- it was his personal copy.  You see I committed the great sin
of not having an SE, and thus never seemed to be able to land the real thing.

Finally at the end of March I had gotten through to the EEP program, and for
a not particularly cheap fee was able to get docs, the 6.304 beta, and the
beta WorkSet Tools.  In the interim, closely following USENET and Compuserve
postings, I did as I was told and purchased OS/2 1.3 in order to get a free
upgrade to 2.0 as this would be the "fastest way" to get a copy of OS/2 2.0 GA.
We spend about $1,000 on beta stuff to this point.

Well fortunately I won a copy at the local roll-out event, or it might have
been another two months before I got one... more on this later.

April 6th or so I sent in my "Proof of License" to get my 1.3 upgraded to 2.0.

Around this same time I found a company "800 software" that not only
knew what OS/2 2.0 was, but even had good enough contacts with IBM to figure
out how to place an order for the WorkSet (for some odd reason the beta
was very restricted and the license expired only 30 days after the GA release.)
The workset cost >$700, and did not include new copies of the programming docs.
I decided to live with the beta copies of the docs, we had spent enough
(I had to upgrade disk and memory to run OS/2, not a biggee, but an expense
none the less.)  God, I hate waiting so long for these shipments!
Hasn't IBM heard of 2-day air?  Why won't the take purchase orders?

Also around this time I finally got the DAP to answer the phone, they sent
me an application, but as I'm kind of in the middle of things by then
I never got around to filling out the forms.  Besides I couldn't get anybody
to tell me how much it costs!  I still have the forms, somebody
want to convince me why I should fill them out?

In late April it all came together, the work was well under way, OS/2 2.0 GA,
while definitely released too soon, was quite usable.  Rumor had it that it
was selling well and the advertising would start any day now, I was happy
because in June or July there would be a CSD that would fix everything, etc,
etc. A free copy of OS/2 2.0 came in, apparently the beta kit entitled you
to one.  The price of the WorkSet went down to $295... oh well, sometimes
you lose.

In May was when I first started having my doubts.  I finally had enough stuff
that I needed TCP/IP access for my PC, 800-Software helped me find the TCP/IP,
and I anxiously awaited it.  I waited and waited and waited, several weeks later
it came in and did not work... no problem -- the net said there was a CSD
so I called the OS/2 tech support line.  They didn't know what TCP/IP was
-- after much fussing around (and many hours lost) I finally got in touch
with the proper people and (after explaining to them how the OS/2 2.0 install
can be aborted for maintenance [bad training... it's a shame there were
eager to learn about OS/2 2.0]), they promptly sent me the CSD, but I would
not be able to use NFS.
During this process, I was harassed a couple of times because I had a clone
with an AT bus, but I said to myself, it will change, IBM is big, it takes
a while to get the word out that clones are not persona non-grata anymore,
the training will eventually catch up, etc.

Time goes on, I still don't have my 1.3 upgrade, I consider calling, but
in my experience that always leads to a bad day(s) and I've got better things
to do.

I had assumed that with the beta tools, that they were slow because of debug
code... Wrong... this compiler's a dog.

In June... finally TCP/IP 1.2.1 -- I really need NFS!!!
I hastily send in my "proofs of license" and expect to WAIT.
Well I did wait, but not that long a week or two and it came -- Yes! things
are finally getting better at IBM.  Oh, but wait they left out my NFS!
Well I'll call them... hmmm.... no number on the invoice... check the original
letter -- no number there either Yikes!!!!  I call the local office
TCP/IP?  what's that?  Who's my SE?  I DON'T HAVE NO STINKING SE!!!!
I yell loud enough somebody gives me the number of the IBM Customer
Satisfaction Number... great this should work.  I call, tell them my
problem (also complain about lack of 2-day air shipment, and also lack of
taking a purchase order).  Yes sir, we'll get back to you on this by tomorrow
afternoon... a week goes by...
Sigh... well there's some nice guys from IBM on USENET...
Finally I get some help.. John Gunther answers my letter.  If IBM gives
awards for service above and beyond this guy deserves one.  Not only
did he Air Express a copy of NFS to me, but he got the Order Fulfillment
Center to cough-up the official copy.

Still I'm greatly disturbed at IBM, they've had enough time to get things
straightened out... is there a Customer Satisfaction Center for the
Customer Satisfaction Center I quip to J. Gunther via e-mail?

Also in June... the projects mostly there... I turn on the optimizer...
Gak!!! it screws up loop expression-extraction optimization.
I loose a week trying to get around the problems... decide I'll just have
to live without optimization... I don't call IBM!!! -- I'd lose another week.

Late June: I put a posting in the OS/2 developers forum on CompuServe
asking about how to get documentation on writing an installable file system
for OS/2 2.0.  I want to write a system that uses a serial line to
remotely access files, and want it well integrated.  Also a friend of mine
has developed a new filesystem technology and wants to explore putting it
into OS/2 2.0.  I still haven't got a response.

Early July.... I decide at the last minute to attend the Windows NT Developer
Conference -- it should be a good platform for our product too, if/when
Microsoft delivers, it should be as good as OS/2, maybe they'll even clean
up some of the grundge that's accumulated in the API's.

July 5th -- Arrive in San Francisco... go to the registration... wow I've
never seen so many nerds in one room... am I at home here or what? :-)
I break out the OS/2 pin I got at the local roll-out and stick it through
my name badge.  Can't have MS think they've got it all!!

July 6th -- The NT Dev Conf begins.  Wow did I ever misjudge NT... it's
clearly superior to OS/2 2.0, and it really exists, they're giving me a copy.
Wow, I'm using on a system that's running NT... God... I only get it to
crash once!  This thing's as stable as OS/2 LA (LA not GA), though the
DOS and 16 bit windows support is still weak... but then they aren't even
trying to come out before December.  My head slumps to the monitor screen...
I remove the OS/2 pin from my name badge.

July 7th -- Dave Cutler speaks first... gives good detail about NT's design
-- wow I didn't know MS could design that solid... amazing.. at mid-day
I decide UNIX is going to lose too.  I call my wife and have her cancel
that copy of OS/2 2.0 ObjectVision we were about to buy, we'll clearly
be running NT at home... and the TCP/IP and file/printer sharing comes for free.
(I have two systems at home, both running OS/2 2.0 and TCP/IP that I paid for
out of my own pocket... sigh.. I guess I just have to take the loss).

July 8th -- I _buy_ a Microsoft tee shirt... In general the prospect of
Microsoft ruling the world makes me nervous but damn it, this product looks
too good too pass up.  And Microsoft is treating developers
(even small ones like me) like kings.  While the conference was $795,
it was well worth the money... I get a copy of NT, several extra CD-ROM with
stuff.  They sign me up for a developer support program automatically where I
get CD-ROMS of helpful stuff occasionally.  The conference is extremely well
planned... Microsoft was very open... listening to what people had to say.
Giving out net addresses of employees.  Good detail on the design of NT.
They're being very careful to say what's there now... what comes in September,
what is future plans... How can I find fault with this?
They were thoughtful enough to _AIR_FREIGHT_ me the 30 pounds of extra
documents so I didn't have to carry them through the airport.
The NT CD has a complete development environment -- C AND C++.
I'll get updates of both docs and CD's up until the commercial release, when
I'll get a copy of all the commercial stuff, but the C/C++ compiler...
what a deal!!!  And my friends and colleagues can get copies too for
only $69/$399 depending on whether or not the want hard copies of the docs.

July 9th -- Arrive home, check mail at work... I've got a letter from IBM
containing the "proof of license" from the 1.3 package saying that I forgot
to check which diskette size I wanted!  Three full months after I send it
in.  Well -- at least the didn't lose it! :-) (I still don't have my
copy of GA from the 1.3 upgrade.)

My product has now entered beta, and some of my testers were noticing
some real slowness problems.  No problem.. I'll run the offending program
through the profiler... wonder what it's called... WHAT! NO PROFILER!!!

Surely you can see my point... were it not for the fact I'm already
in beta I would have to very seriously consider switching to doing a version
for NT only... How many other developers are facing the same situation... plenty
I'm sure.  It's not that OS/2 2.0 is bad, it's that IBM bureaucracy sucks!

What I as a developer want from IBM:
	IBM to start providing information on OS/2 in a timely open manner.
		Questions on Compuserve should always be answered, and in
		a timely manner.  Generalized sales figures should be released,
		users and developers should know where things stand.  If they
		are not released, then IBM shouldn't complain if people assume
		the worst.
	IBM should have provided a good, working, tested development
		environment with C and C++ at at a reasonable price, from day
		one... WITH ALL THE PROGRAMMING DOCS (at least on-line), and
		the hard copies of docs AT COST.
		It simply can't be in IBM's interest to try to make money
		off of documents for developers, it only leads to people
		making due with as little as possible.
		I want a PROFILER, and I want it two weeks ago!
		I want a working optimizer two months ago.
	RESPONSIVE customer/developer support.
	IBM to start shipping (possibly for a small extra fee) by 2-day air.
	IBM to accept purchase orders.
	Super-VGA (and others) drivers... it doesn't matter that they have to
		be written by the card manufactures...  IBM
		should have provided proper encouragement IN ADVANCE!
		The NT pre-beta is shipping with more drivers than what comes
		with OS/2 2.0 GA.
		I am very tired of 640x480 16.
	IBM to start promoting OS/2... PS/2 television commercial and magazine
		advertising is not OS/2 advertising.  I have never owned
		a PS/2, few people I know do (systems that will be obsolete
		in 3 years don't have to be built like
		presidential bomb shelters), and I probably will never own
		a PS/2... that doesn't mean I don't like OS/2.
		Emphasizing PS/2s over OS/2 will simply kill both.
	IBM to either give me an SE, or quit asking me who he/she is!
	IBM to get that "June" CSD out the door -- working and complete.
	IBM to somehow get tutorial documentation for 2.0 published.
		Where are the third party books?  I swear it's easier to
		find out about NT that it is OS/2, and NT isn't even
		a product yet.  Does IBM not have an equivalent of MS-PRESS?
	IBM to present their plans openly.  In this month's Computer Shopper
		there is an article saying that IBM has said that they are
		deemphasizing OS/2 for Pink and that full OS/2 - Pink
		integration will occur later... perhaps in 1995.
		If this is true, and I've been mislead about the positioning
		of OS/2 I might develop for the Apple-side of Pink... but
		IBM can count me out.  I can tell you (with large grains
		of salt being necessary of course) where Microsoft will be
		for the next 2 years.  I have no idea where OS/2 will be.
		I simultaneously see IBM saying that OS/2 will be portable
		and have symmetric multiprocessing by November... yet another
		part saying as of two months ago it was not decided if OS/2 3.0
		will be based on NT or Mach (now it's Mach I'm told).
		It is impossible for any company that does not know which such
		OS they're using as a base in May, to have a delivered product
		in November.

Sincerely,

Brian Sturgill
(801) 581-5591
brian@cs.utah.edu  (Internet address, my preferred contact point)
CompuServe: 70363,1373 (personal)
CompuServe: 76300,3115 (U of U)
(Spin off company account setup still in progress)