[COFF] Joys of PL/I [Was: Re: [TUHS] Book Recommendation]

[Charles H Sauer]

On 11/16/2021 12:54 PM, Nemo Nusquam wrote:
> On 2021-11-16 09:57, Douglas McIlroy wrote:
>> The following remark stirred old memories. Apologies for straying off
>> the path of TUHS.

>>> I have gotten the impression that [PL/I] was a language that was 
>>> beloved by no one.
>> As I was a designer of PL/I, an implementer of EPL (the preliminary
>> PL/I compiler used to build Multics), and author of the first PL/I
>> program to appear in the ACM Collected Algorithms, it's a bit hard to
>> admit that PL/I was "insignificant". I'm proud, though, of having
>> conceived the SIGNAL statement, which pioneered exception handling,
>> and the USES and SETS attributes, which unfortunately sank into
>> oblivion. I also spurred Bud Lawson to invent -> for pointer-chasing.
>> The former notation C(B(A)) became A->B->C. This was PL/I's gift to C.

>> After the ACM program I never wrote another line of PL/I.
>> Gratification finally came forty years on when I met a retired
>> programmer who, unaware of my PL/I connection, volunteered that she
>> had loved PL/I above all other programming languages.

> My first language was actually PL/C  (and the computer centre did not 
> charge for runs in PL/C).  I needed to use PL/I for some thesis-related 
> work and ran into the JLC wall -- no issues with the former, many issues 
> with the latter.  One of the support people, upon learning that I was 
> using PL/I, said: "PL/I's alright!"

Inside IBM in the 70s, PL/I was definitely not "insignificant". From my 
perspective it was the most reasonable language available on VM/370. I 
arrived at Yorktown in 1975 fresh from Austin with a couple of boxes of 
punch cards of my "APLOMB" simulation program written in Fortran. I was 
surprised when I was encouraged to pursue major enhancement of APLOMB 
and dismayed by continuing in Fortran. After a period of increasing 
frustration, I wrote a SNOBOL program to convert APLOMB to PL/I and 
become the basis for RESQ 
(https://notes.technologists.com/notes/2020/08/25/remembering-resq/). As 
long as I avoided questionable parts of IBM's PL/I, I was happy with 
PL/I. It is hard to imagine that RESQ would have succeeded in any other 
language.

-- 
voice: +1.512.784.7526       e-mail: sauer at technologists.com
fax: +1.512.346.5240         Web: https://technologists.com/sauer/
Facebook/Google/Twitter: CharlesHSauer