Ancient C compilers

Stackoverflow, while a really great site, sometimes has a SPAM policy that is a bit too strict. Of course, it is logically they want to prevent people from advocating their own software. A question that would initiate a lot of answers regarding pointers to different ancient C compilers and would be a great resource for people online, can not make it.

Why bother about old code?

Links to old code give at times quite some authority to a question. See this reference to an analysis of source code of an early compiler, e.g.:

• Why was the switch statement designed to need a break? where is referred to the Sun ANSI C compiler front end by Peter van der Linden.

If you really want to know why switch statements have breaks or why other choices have been made, it might be worthwhile to look how they were used in the code. For example, I see a lot of fall-throughs in this code! Off-topic, but I think the real reason is different. A switch statement is quite a "flat" structure. A person who likes a switch statement does probably not like a piece of code like this (with implicit breaks):

switch(p) {
case 0: 
	if (error_cond1) {
		// do something
	} else {
		if (error_cond2) {
			// do something
		} else {
			// do something
		}
	}
case 1:
	// do something
}

But probably more like this:

switch(p) {
case 0:
	if (error_cond1) {
		break;
	} 
	if (error_cond2) {
		break;
	} 
	// do something
	break;
case 1:
	// do something
	break;	
}

Just my two cents, but I personally understand why people who love switches would love breaks as well, especially if the horizontal screen estate was much more limited in those days.

A quick pcregrep -Mi 'case.*\n.*case' * | wc -l in the c directory shows 166 fall-throughs by the way, and these are only subsequent case statements on subsequent lines. I would say, that's quite a lot on a total of grep -ri case . | wc -l: 459.

Source code

References to source code of early compilers:

• Ritchie describes a DECtape found by Paul Vixie and Keith Bostic. Which can be found on github elsewhere as well as legacy-cc. To honor Ritchie it is here added in his words as "primeval C".
• This code is by a combined effort from others combined to a running system at google code, and put here in the folder [unix_v1].
• Sixth edition of Unix by Bell laboratories released in May 1975, which is designed to run on the PDP-11. The files are from Dennis Ritchie. The compiler can be found in the [unix_v6] folder and then the c/ directory.

Todo

Cardeva made a lot of source available on the Unix Archive, e.g. see the minnie.tush.org mirror. Interesting code needs to be extracted from there.

Documentation

• https://en.wikipedia.org/wiki/Portable_C_Compiler by Stephen C. Johnson of Bell labs in mid seventies
• https://en.wikipedia.org/wiki/Amsterdam_Compiler_Kit by Andrew Tanenbaum in early eighties (but not open-source of course, or we wouldn't have had gcc)
• https://en.wikipedia.org/wiki/GNU_Compiler_Collection by Richard Stallman in 1987