Pascal not considered harmful - was Re: Rich kids are into COBOL

Mark Green markwgreen at
Thu Feb 19 20:33:12 CST 2015

I started using Pascal on CDC 6000 series machines.  At that point in time
the other languages that were widely available on CDC hardware were Fortran
and assembler.  Within a year or so we dropped assembler as our main systems
language and switched to Pascal.  Pretty much everything we could do in
assembler we could do in Pascal in fewer lines of source code.  Writing
efficient assembler on CDC machines was not particularly easy and the Pascal
compiler generated fairly good code, so we saw little performance

With the mixture of 6 bit and 12 bit character codes (plus a few others)
along with the bizarre end of line conventions, Pascal code tended to be
easier to understand and maintain than assembler.

Wirth had a pretty good understanding of the CDC architecture having written
several compilers for it before Pascal.  Pascal was tuned to that
architecture and the CDC operating systems.  Some of the features that
people not used to CDC machines find strange were aimed at the shortcomings
of the CDC machines.  In many ways Pascal on the CDC machines was similar to
C on PDP11s.  I worked on both for several years and found switching back
and forth to be relatively easy.

On the CDC machines Pascal was far from academic and useless, particularly
in practice.  You need to understand the machine architecture and system
environment where a language was developed to judge whether it was practical
or not.

-----Original Message-----
From: cctalk [mailto:cctalk-bounces at] On Behalf Of Tor Arntsen
Sent: February-19-15 7:20 PM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Pascal not considered harmful - was Re: Rich kids are into

On 19 February 2015 at 17:40, geneb <geneb at> wrote:
> I suspect it was Borland's extensions to Pascal that removed any 
> limitation in I/O.

That's right.   But there's more than I/O. The academic-tool variant
of Pascal, as Wirth designed it, was simply useless in practice, or
extremely cumbersome to use because you couldn't design a function which
could take arrays of variable sizes as input, you had to declare one
function for each size. Hopeless. You couldn't do any real data processing
that way. Turbo Pascal, and every other useful variant, e.g. the Pascal I
used on a minicomputer, fixed that part, and often added I/O extensions in
various ways. In short, they made the language flexible, and thus usable.

Then TP of course had that fast edit-compile-execute cycle, a low price, and
the super-easy IDE. The learning curve from getting your hands on TP to
actually use it was very low. Actually the Turbo Pascal IDE is still the
only IDE I like. I don't use any of the modern ones, they are just in the
way. But I recently tried the CP/M TP version again, haven't used it since
the eighties.. and I still like the IDE.
Re-learned it in seconds.

More information about the cctalk mailing list