Some of the error messages produced by Apple's MPW C compiler (2006) (opens in new tab)

    void outchar (char c) { 
        c = c | 0x80;
        asm volatile ("jsr $fbfd\n" : : "a" (c): "a");
    }

    void outstr (char* str) {
        while (*str != 0) 
            outchar(*str++);
    }

    void main () {
        outstr("Hello, world!\n");
    }

    lda #$c8       ; ASCII H | 0x80
    jsr $fbfd
    lda #$e5       ; ASCII e | 0x80
    jsr $fbfd
    ...

[1] https://llvm-mos.org/wiki/Welcome

For some reason everyone prefers the newer software, though. Perhaps there’s more to it than binary size?

http://telcontar.net/Screenshots/worldwidewonk/Amazon-404-eh...

mastry found a screenshot of the image in his sibling reply.

0 - http://g-ecx.images-amazon.com/images/G/01/x-locale/common/k...

1 - https://web.archive.org/web/20071030172825/http://www.amazon...

Another anecdote: somewhere in that time period we (Prime) were using comments for various metadata and pre-release and post-deployment checks would verify the comments were in place. The team responsible for the edge proxies decided they were going to rewrite all the HTML going out on the fly to remove extraneous comments and whitespace in an effort to reduce page-size.

In the middle of testing a release all of the tests related to a particular feature started failing and (I believe) different devs were getting different HTML on their systems (the feature wasn't rolled out to every session). Our QA team was extremely pedantic in the best way possible and wouldn't allow the release to continue until testing could complete, so we had to track down the responsible parties and get them to dial down their transformation. They eventually tried again without stripping comments, but I can't imagine much was saved after compression without much of anything removed (they might have been rewriting other tags as well).

AFAICT, kids own the copyright to things they create[0], but guardians are responsible and can use it on the child’s interest. IANAL, consult an attorney, etc., etc.

0 - https://www.copyright.gov/help/faq/faq-who.html#:~:text=Can%....

Regarding "everything had to fit in RAM": prior to real virtual memory, the Macintosh Resource Manager was capable of loading and unloading resources on the fly. Resources marked purgeable could be discarded when memory was needed. Code segments (another type of resource) could be loaded by automatically by the Segment Manager, but as you said would not unload until the application requested it or exited. INITs (system extensions) unloaded all code after initialisation by default (requiring extra steps to keep anything in RAM).

Virtual memory was built-in by System 7 (and I think available on supported hardware via 3rd party utilities earlier?).

Dereferencing a void pointer has no meaning. The compiler can do anything it wants because it doesn't know how to interpret the memory. It could give you the correct thing, it could warp a civilization in from a distant planet, or it could open a world-ending black hole. All are equally probable.

It's super-useful to temporarily comment out a bit of code, and then to comment out a larger block surrounding it. Especially when debugging.

Sadly I've never used a language that supported that.

e.g. '/*' and '*/' would match each other, '/**' and '**/' would match, and so on.

That way, you would have full control of the depth of the comments, removing other comments wouldn't break the inner comments, etc.

I do run into the same issue you're describing, so I think there's value in the idea.