Every few years I go and dig out a 65-line C program that I wrote as a grad student and try to gently poke it through a compiler but I finally got tired of that and replaced it with a 10-line awk script.
Wow, take a look at this jalopy! Written ca. 1977 (with ed(1), the standard editor) on a PDP-11/45 running V7, prob'ly on an ADM-3A "glass terminal" in the terminal room on the second floor of the E-Quad.
![#include <stdio.h>
int print = 0;
main(argc, argv)
register char **argv;
{
register short printname;
if (--argc && argv[1][0] == '-') {
print++;
--argc;
argv++;
}
if (argc == 0) {
Long();
exit(0);
}
printname = argc > 1;
while (argv++, argc--) {
if (freopen(*argv, "r", stdin) == NULL) {
perror(*argv);
continue;
}
if (printname)
printf("%s: ", *argv);
Long();
}
exit(0);
}
Long()
{
char tmp1[512], tmp2[512], *in = tmp1, *save = tmp2, *hold;
register int max = 0, i = 0, c;
while ((c = getchar()) != EOF) {
if ((in[i] = c) == '\n') {
in[i] = '\0';
if (++i > max) {
max = i;
hold = in;
in = save;
save = hold;
}
i = 0;
} else if (i == 511) {
in[i] = '\0';
while ((c = getchar()) != '\n' && c != EOF)
i++;
if (++i > max) {
max = i;
hold = in;
in = save;
save = hold;
}
i = 0;
} else
i++;
}
printf("%d\n", max);
if (print)
printf("%s\n", save);
}](https://files.a2mi.social/media_attachments/files/114/654/262/701/655/224/original/efd89f1cf5769ca1.png)
@peterhoneyman I remember you mentioning this program to me at one point while walking through the CSE building…16 years ago. Since then, I’ve occasionally thought about it.
@jeffpc I challenged Claude to rewrite it in awk and it came up with the same script that I wrote by hand.
Not sure what this means.
@jeffpc then i asked claude to convert that awk script to c and it introduced a buffer overflow lol
@peterhoneyman but did it use the 'register' keyword to make sure it is really fast at overflowing?
(BTW, I am amused by your usage of 'register' on argv. I know nothing about the calling conventions in V7 on PDP-11, so that smells like a silly attempt at an optimization that probably didn't accomplish anything measurable given that this program does I/O.)
@jeffpc yes, it is amusing in retrospect. iirc, twenty-something me had certain beliefs about the architecture he was dealing with, like “you have three general purpose registers available, use ‘em or lose ‘em.” obviously, the compiler can do a much better job allocating registers than some skinny long-haired chain smoking grad student lol. also, as you observe, this program will run at approximately the speed of cat(1) no matter what.
@peterhoneyman "use 'em or lose 'em" would be a cruel compiler code generation behavior :D