10 May
2024
10 May
'24
12:10 a.m.
On May 9, 2024, at 7:55 PM, Fred Cisin via cctalk
<cctalk(a)classiccmp.org> wrote:
True, and Stroustrup added that "and C++ is a cannon that blows off your entire
leg".
> ...
> I've written code in Pascal, as well as Modula-2. Never liked
> it--seemed to be a bit awkward for the low-level stuff that I was doing.
On Thu, 9 May 2024, Paul Koning via cctalk wrote:
Not surprising, since that's not what it is
all about. Both, like their predecessor ALGOL-60 as well as successors like Ada, are
strongly typed languages where doing unsafe stuff is made very hard. Contrast that with
C, which sets out to make it easy to do unsafe things and partly for that reason has a
feeble type system. So doing low level stuff like device drivers is difficult, unless you
create extensions to break out of the type system. An example of how to do that is the
Burroughs extension of ALGOL called ESPOL, which is what they used to write the OS.
Actually, Burroughs did a number of extended versions for different purposes; there's
also DCALGOL (Data comm ALGOL) intended for writing communications software. Why
that's separate from ESPOL I don't really know; I only ever got to do regular
ALGOL programming on Burroughs mainframes. One reason for that: those systems depend on
the compilers for their security; if ordinary users got access to ESPOL they could write
dangerous code, but in ALGOL they cannot.
One of the things that _I_ love about C is that it is easy to get it out of the way when
you want to do something lower level.
Rather than feeble type system, it could have had a requirement to explicitly
"cast" anything being used as a "wrong" type.
One of Alan Holub's books about C is titled
"Enough rope to shoot yourself in the foot" Each language has its own specialty. And you need to
find the one that fits you best.
It used to be (and likely still is), that every computer science grad student created a
new language. A requirement (usually UNSPOKEN) was that the compiler be able to compile
itself. That the language compiler is written (actually normally RE-written) in that
language and compiled by that compiler. That certainly seems to bias things towards
languages that are well suited for writing compilers! If you were to create a language
that was specializzed for something completely different, and poorly suited for writing
compilers, then it would not be respected.
If you don't mind the total lack of protection, FORTH is very nice: it even more
easily than C lets you do low level things, and it is also very small. And the
implementation is by definition entirely extensible. A large FORTH program I wrote in the
1980s, on PDP-11 FORTH, starts out by redefining the language as a 32-bit version.
I still remember a classmate of mine, who told me when we were both at DEC that he had
written an expression parser in COBOL. I think he also tried to do one in RPG but found
it was too hard.
paul