Re: call by name, was silly question: prefix vs postfix ops

William Clodius <wclodius@los-alamos.net> wrote:


<snip>
> FWIW a year or two ago I glanced through Wexelblatt (History of
> Programming Languages) at the local library, and while Peter Naur in
> his presentation said he thought only one person on the committee
> understood the implications of pass by name, other European members
> disagreed with him. Neither Perlis, who introduced Naur, nor Backus
> who should have been at the talk given he gave the Fortran talk at the
> same symposium, made recorded comments on that topic.
>
> [Perlis told me that call by name was a mistake. They were trying to
> do an elegant definition of Fortran-style call by reference, and were
> rather surprised when Jensen pointed out what they'd done. -John]


I know. You have cited this several times over the years. What I meant
to imply, but should have stated explicitly was


1. Peter Naur implied in his talk during the (first) History of
Programming Languages symposium that he roughly agreed with the comments
that Perlis made to you, i.e., that the committee did not understand
call by name.


  2. In the discussion afterwards, the other two European members of the
Algol 60 committee insisted that they did understand the implications
of call by name when it was adopted. Their language indicates that they
were angry at the implication that they did not understand it at the
time.


FWIW the symposium occurred in 1978, although the edited proceedings
were only published in 1981. If that was after the time Perlis made his
comments to you, he might have changed his opinion.


[Perlis would have told me that around 1980 when I was his grad
student. If other members of the committee say they realized that
they were designing a language where procedure arguments are in effect
macros that have to be reexpanded on each refrerence via a callback, I
have little basis to disagee, but it does seem odd that as best I can
tell, nobody said anything about the complexity of implementing call
by name until Jensen's paper. -John]

Return to the comp.compilers page.
Search the comp.compilers archives again.