From: firstname.lastname@example.org (Robert Corbett)
Subject: Re: Labels .. Pros and Cons
Date: 28 Apr 1999 03:14:17 GMT
In article <email@example.com>,
Richard Maine <firstname.lastname@example.org> wrote:
>Clive Page <email@example.com> writes:
>> I've read and enjoyed the "COME FROM" paper, and heard rumours of an actual
>> implementation: can you reveal the secret of who did it?
>It was implemented in the Elxsi Fortran compiler, sometime in the late
>'80s. Bob Corbett was then working for Elxsi (he currently works for
>Sun and posts here occasionally). I recall that Bob suggested,
>without saying what I'd find, that I use the debugger to peruse an
>area of the compiler executable. Turned out to be an area with
>strings with the names of a lot of familliar-looking Fortran
>statements. The string "comefrom" appeared there.
>I can verify first-hand that it worked because I wrote a trivial
>program to test it, making the obvious guess of its syntax and
>meaning. It was, of course, undocumented, and I hadn't actually read
>the paper at the time. Rumor has it that it was removed from the
>compiler after someone (not me) submitted a bug report about comefrom
>loops not optimizing as well as goto loops.
Yes, ELXSI Fortran implemented both the COME FROM and the conditional
COME FROM statement. They were implemented by Ralph Merkle of trapdoor
knapsack and nanotech fame. I don't recall that the COME FROM statement
was ever deleted, although ELXSI Fortran effectively has been.