[J3] [BULK] Re: Niklaus Wirth

[Van Snyder]

On Mon, 2024-01-08 at 20:23 +0100, Toon Moene via J3 wrote:
> Of course, all that programming we physics 
> students would do would be in Fortran, but the "lab" work in "computer 
> science" was in an interpreted version of Pascal.

The IBM 1401 FORTRAN-II compiler produced interpreted code for floating
point. It was actually quite well done, with decimal arithmetic from
six to 20 digits, and two guard digits, but no symmetric rounding.

I have the source code if anybody wants it. It was a single pass
compiler that read the source code into high memory, then loaded 63
phases below it to gradually massage it into executable code. A bit of
a miracle on a 16k machine. Few of the phases had more than 300
instructions. There was an option to punch an object deck, so you
didn't need to load and run the 2,200-card compiler every time. After I
got it going I ran one of the demos from the manual and it got the
wrong answers. There was a bug in the divide code in the SimH emulator.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20240108/7b336628/attachment-0001.htm>