Re: A database theory resource - ideas
From: David Cressey <cressey73_at_verizon.net>
Date: Tue, 20 Mar 2007 12:41:48 GMT
Message-ID: <gEQLh.9499$DX5.3224_at_trndny06>
> On Mar 19, 2:17 am, "Tony D" <tonyisyour..._at_netscape.net> wrote:
> > On Mar 18, 1:53 am, "Marshall" <marshall.spi..._at_gmail.com> wrote:
> >
> > (Another one down the Google Groups memory hole. sigh...)
> >
> > > Oh, pooh. The problem isn't Kernighan, nor Ritchie, nor C itself.
> > > There are perfectly good uses for a low-level programming
> > > language. The problem is the legions who came afterwards
> > > who didn't recognize C for what it was, and instead turned
> > > it in to the One True Way.
> >
> > No, it's worse than that. Aside from C being a macro assembler
> > pretending to be a high level language,
>
> Ah! But C does not pretend to be high-level, nor were K&R claiming
> it was. In fact, the very book under discussion says "C is a
> low-level programming language ..." Again, I blame the legions
> coming afterwards. (What's the opposite of "vanguard?")
>
>
> > the exposition of it in book
> > form, "The C Programming Language", is a woeful attempt at describing
> > it. It has no redeeming features - it isn't short (compared to the
> > Algol-60 Report or the Pascal Report, which also do a better job
> > describing the respective languages), and it is insufficiently formal
> > (apocryphally, you could write 4 compilers which all agreed with some
> > interpretation of the text but which all produced different results
> > with the same source code - since they all "agreed" with the text,
> > they were all "correct"). Both of these are fatal flaws in any kind of
> > specification, and fatal to the max in a language spec. Sadly, others
> > have been influenced by both the language and the style of exposition.
>
> It would be culturally insensitive* to judge the book and the language
> by current standards. You fault them for the lack of formality, but
> when did formal methods for describing programming languages
> really get started? Scott and Strachey first began working together
> in 1972, the same year "The C Programming Laguage" was
> published. So we can't fault K&R for not using denotational
> semantics, can we? What *was* the first language with a
> formal semantics? Offhand I'd guess ML, which was when?
> ... It seems "The Definition of Standard ML" came in 1990,
> fully 18 years after TCPL.
>
Date: Tue, 20 Mar 2007 12:41:48 GMT
Message-ID: <gEQLh.9499$DX5.3224_at_trndny06>
"Marshall" <marshall.spight_at_gmail.com> wrote in message
news:1174361685.002881.173570_at_b75g2000hsg.googlegroups.com...
> On Mar 19, 2:17 am, "Tony D" <tonyisyour..._at_netscape.net> wrote:
> > On Mar 18, 1:53 am, "Marshall" <marshall.spi..._at_gmail.com> wrote:
> >
> > (Another one down the Google Groups memory hole. sigh...)
> >
> > > Oh, pooh. The problem isn't Kernighan, nor Ritchie, nor C itself.
> > > There are perfectly good uses for a low-level programming
> > > language. The problem is the legions who came afterwards
> > > who didn't recognize C for what it was, and instead turned
> > > it in to the One True Way.
> >
> > No, it's worse than that. Aside from C being a macro assembler
> > pretending to be a high level language,
>
> Ah! But C does not pretend to be high-level, nor were K&R claiming
> it was. In fact, the very book under discussion says "C is a
> low-level programming language ..." Again, I blame the legions
> coming afterwards. (What's the opposite of "vanguard?")
>
>
> > the exposition of it in book
> > form, "The C Programming Language", is a woeful attempt at describing
> > it. It has no redeeming features - it isn't short (compared to the
> > Algol-60 Report or the Pascal Report, which also do a better job
> > describing the respective languages), and it is insufficiently formal
> > (apocryphally, you could write 4 compilers which all agreed with some
> > interpretation of the text but which all produced different results
> > with the same source code - since they all "agreed" with the text,
> > they were all "correct"). Both of these are fatal flaws in any kind of
> > specification, and fatal to the max in a language spec. Sadly, others
> > have been influenced by both the language and the style of exposition.
>
> It would be culturally insensitive* to judge the book and the language
> by current standards. You fault them for the lack of formality, but
> when did formal methods for describing programming languages
> really get started? Scott and Strachey first began working together
> in 1972, the same year "The C Programming Laguage" was
> published. So we can't fault K&R for not using denotational
> semantics, can we? What *was* the first language with a
> formal semantics? Offhand I'd guess ML, which was when?
> ... It seems "The Definition of Standard ML" came in 1990,
> fully 18 years after TCPL.
>
What about Algol-60? What about Pascal? These were described in fairly formal terms by 1972.
Still, I don't fault K&R for the lack of a formal description in 1972. I don't think their intended readership was generally those people who would have benefitted from a formal presentation. Their goals were to get a minimalist OS and PL off the ground with minimal resources, both people and computers. They succeeded.
It's the people who came later, as you suggest, that expanded the scope, while blurring the original vision. Received on Tue Mar 20 2007 - 13:41:48 CET