On Sunday, 96/08/18, Murray Lesser wrote to David Noon about "PL/I for
OS/2" as follows:

ML>     I don't know the time lag on DevCon deliveries between here and
ML> there, but the "SERVICES" directory file on DevCon 10, Disc 1 (the
ML> one with the PL/I CSDs in it) is dated late March, and I think I got
ML> the volume in late May or early June.  (DevCon 11 should arrive any
ML> day now!)

Hi Murray,

The word on CIS was that first week of September should see Dev Con
11. We get them about a week later, from IBM Denmark.

ML> My FTPed files of Personal Edition CSD#7 are also dated
ML> late March; I assume the issue date for the Professional Edition was
ML> probably slightly earlier, but still may have been too late to make
ML> the DevCon 10 deadline, seeing as how it takes so long for one IBM
ML> hand to find out what another one is doing!

But 3 months is a long time. The Professional Edition CSD's are
usually about a week or 10 days ahead of Personal Edition (corporate
customers still have all the clout), so it is even longer.

ML> It is my understanding
ML> that CSD#3 for the toolkit is the one that brings it up to compiler
ML> release 1.2.  Have you already upgraded?

I haven't upgraded yet. The upgrade is US$449, so in this country it
will probably be about 450 pounds plus VAT, so call it US$850.

Hmm ... if I wrote a program to win the football pools (soccer season
has just started) I could ... probably do better pricking the entry
form with a pin. But to cover a cost of that magnitude I shall need to
make some budgetary adjustments.

ML>     Last communique from Carolyn said no decision yet on "promotional
ML> price" for upgrades to 1.2.  (Stay tuned!)  Perhaps I'll wait for
ML> fall. 

The word on CIS late last week was that promotional pricing had been
nixed by product management. I guess they saw a reduction in support
costs if fewer people bought the product. Perhaps that's IBM's "OS/2
strategy" from now on. ... :-(

ML> Release 1.1 seems to work fine for what I am doing these days,
ML> and I suppose my utilities can always be recompiled and relinked
ML> with 1.2 when and if.

When you see the enhanced tools it will be when, not if. [Assuming the
OS/2 1.2 has the same profiler, etc., as the NT 1.2.]

ML> Actually, I may be better off staying with
ML> 1.1!  Switching to 1.2 will mean I have to provide separate
ML> libraries to be called by applications with and without
ML> multithreading, something not required with release 1.1.

Everything is thread-safe in 1.1, but the single-threaded library has
some perhaps leaner, definitely non-reentrant, code. The beta NT
version sizzled.

However, you should be able to supply the lot in one bundle. The DLL's
should have separate names.

ML> (Shades of
ML> MS BASIC PDS, which required a separate library of
ML> separately-compiled (or assembled) procedures for each mode the
ML> compiler was capable of--even though the source-code for each was
ML> identical!)

Bleah! Memory models. ... :-(

ML>     At the moment, I am contemplating "porting" the full-screen
ML> text-mode "windowing" subroutine library that I wrote for BASIC
[snip]
ML> otherwise), making VIO calls to replace the assembled screen stuff,

Are you going to try the 32-bit VIO that has been on Dev Con for the
last year or two?

You could try your hand at macro writing by having a single "switch"
to change the decalrations of the VIO calls between the 16=bit and
32-bit versions. [Also remember that using BYADDR was a foolproof way
of choosing between PTR and PTR SEGMENTED to pass parameters by
location.]

ML>     To paraphrase from the hand-wringers on the OS2 conference:  Usual
ML> poor IBM marketing.  Perhaps you just didn't proselyte enough, and
ML> I am the only person who ever bought the Personal Edition. :-(

Well PL/I was a special effort, by refusing to allow software
resellers to market this "hardware device" for them. Incompetence
cubed.

Regards

Dave


 * KWQ/2 1.2i * Gross incompetence (n.): 144 COBOL programmers
--- Maximus/2 3.01