On Thursday, 96/01/11, Murray Lesser wrote to David Noon about "Pl/i
Miscellanea" as follows:

ML>     I tried ILINK.  The FREEFORMAT (default) mode is nice after one
ML> figures out how to use it ( INF manuals!!).  I
ML> got slightly smaller executables than with LINK386--even though I
ML> didn't try the OPTIMIZE option.  I wonder if there is any way to get
ML> CSet++'s ICC to call ILINK instead of LINK386.

Hi Murray,

I haven't tried ILINK yet, but I plan to this weekend. I am
contemplating switching from WorkFrame/2 V1R1 to WF/2 V2R1 (plus much
maintenance), so I might set it up in the new environment.

I think C Set ++ has LINK386 options hard-coded, so getting it to use
ILINK could be a challenge. However, you can always edit a makefile for
C Set ++.

ML>     Do you think the three new PLITK add-ons (LPEX, ILINK, and ILIB)
ML> will ever show up on DevCon, or will they remain private to us
ML> PL/I aficionados?  It seems that LPEX will parse-edit C/C++, Pascal,
ML> etc., as well as several mainframe languages I've never heard of.

I doubt they'll be on the DevCon soon, but they are included in Visual
Age C++ and COBOL. Indeed, the reason for ILINK's existance is to try
to cut down the code bloat resulting from large VFT's in C++.

LPEX is rather a nice editor for a bundled freebee. I have been
playing with the REXX scripts to customise it further. It should
ensure that EPM will never be sold, since LPEX is way better than EPM
for a programmer's editor.

ML>     None of the (Personal Edition) IBMPLI\INCLUDE files has been updated
ML> since January, 1995.  According to the readme with PLITK CSD#2,
ML> nothing in PLITK itself has been updated since CSD#1.

The same applies to Professional Edition too. I am planning to rewrite
all that stuff to encapsulate the OS/2 API inside PL/I statements. I
have given up hope of decent %include files from IBM.

ML> Most of the
ML> DLLs in IBMPLI\DLL were updated in December 1995, including a new
ML> run-time DLL to be distributed with programs compiled with CSD#6. 
ML> (Wonder if these new run-times are retroactive; I thoughtlessly
ML> recompiled my few PL/I utilities before I thought to run a test.)

If it is anything like the mainframe compiler and library, the new
dynamic run-time will support old object code without recompiling or
relinking.

ML>     Thanks for the tip and the warning.  (I don't think I would ever
ML> have figured out what UNSPEC() really does from the entry in the
ML> manual!)

It is intended to remain hidden until it is a last resort. One should
always regard UNSPEC() as a last resort. (Well, second last, if you
count calling assembler code.)

ML> I thought PL/I was the exception to the rule that
ML> computer languages must include built-in booby traps [e.g.,
ML> sprintf() in C], but it seems not.

When one analyses someone else's PL/I, the first thing to check is
excessive use of UNSPEC(). It is a double-edged implement of the
sharpest kind.

ML>     Sorry, but I missed your DOS incarnation.  Have I misread you, or
ML> are you saying that there is (was) a PL/I for DOS?  If so, is it
ML> still available?  I run my ancient laptop on PC-DOS 5/DESQview and
ML> it would be nice to have PL/I for it.  Maybe I should ask Carolyn.

There was, and might still be, a PL/I for DOS, and another for CP/M.
They were written by Dr. Gary Kildall, who wrote CP/M, and published
by his company, Digital Research Inc. Unfortunately, DRI was bought by
Novell about 2 years ago, so I doubt you could get a copy now.

I have asked Paige Vinall about such a compiler coming from IBM and
she said there would be no new 16-bit compilers from IBM. However, a
PL/I for Windows NT is going (or is expected to go) into beta in a
couple of weeks time.

Regards

Dave


 * KWQ/2 1.2i * My strength is as the strength of ten because my code is pure.
--- Maximus/2 3.01