Just for some context, the MPW C compiler that produced those messages was actually not developed internally at Apple, but was rather done by Green Hills Software [1] under contract as mentioned on the wikipedia page [2] and its source [3] which is funnily enough about this exact same topic.
I forget what happened to the Green Hills folks, but the official MPW C compiler was an offshoot of the MPW Pascal compiler, and was written by Roger Lawrence at Apple in about a year (around 1987-1988) largely as a reaction to what he thought was a terrible contract.
The Green Hills folks were likewise upset at Roger's effort. "Anyone who says they can write a compiler in a year should be fired." Roger himself held the same opinion, a couple of years later :-)
I sat next to Roger while he wrote that thing, he was in the next cubie over (I was working on the MPW linker and some other tools). He types really fast.
Hello from a guy who used your and Roger's work every day back then. Still one of my favorite developer experiences. (I ported an old-school AI application from its original Sun workstation environment to the Macintosh. Had a blast but our product crashed and burned with the AI winter).
> Anyone who says they can write a compiler in a year should be fired." Roger himself held the same opinion, a couple of years later
Nonsense.
Three of us did it in a year compiling a language similar to PL/M written in Bliss 36 on a PDP-10. And that was when compilers were much harder to do then than they are now.
Well, they were clearly upset at losing the contract.
But mostly they were trying to say, "Okay, you probably have a guy who can do that. But it will be a buggy mess and it will go better if you don't give that dev permission to be a hero."
I still don't follow. A year seems like plenty of time to build a c compiler. Maybe it won't generate the best code, but no compiler did, back in the day.
Sounds like GH's founder just really likes dinosaurs. (To be clear, if I was in that kind of financial position, I'd absolutely have a T-rex skull, or better yet the whole dinosaur, on display in the lobby).
I think my first real push with MPW was using it as an environment to host the AT&T C++ compiler, which at that time was still a bunch of preprocessor macros written by Bjarne himself.
I was introduced to MPW when I started at Apple in 1995. I started on Quickdraw GX (if anyone remembers that).
Not specifically related to MPW, more about the times, a clean compile of the Quickdraw GX framework took something like eight hours. (Fortunately incremental builds were much quicker.) Nonetheless, it was common to kick off a clean build as you were leaving the office for the night.
Funny to think about that compiler compiling all night as I had dinner, watched a little cable TV, got some sleep....
I was a cfront beta site using MPW; for people that don't know, that was the self-compiling C++ from C, before the first AT&T C++ compiler, in the late 1980s.
I agree and we are saying the same thing. There was no C++ compiler, the C++ language was processed by cfront, similar to the preprocessor and include compile phase. AT&T supplied cfront. The language spec was changing. I wrote applications with it, not checking "language lawyer" syntax problems. I wrote general feedback on performance, and things that would make the language practical. It is hard to describe those days, but concepts of object-oriented programming were new to many people, and I was quite interested in building actual applications. The linker was the common linker for multiple other languages, like gcc today.
As long as it doesn’t contravene the Apple Style Guide:
“Humor can enhance documentation by adding to a reader’s enjoyment and by helping to lighten the tone. Humor usually works best in examples, where it’s less likely to distract the reader. Be careful that your humor is in good taste—one reader’s joke can be another reader’s insult—and keep in mind that humor may not translate well in localized text.”
It blows my mind how common taking a silly tone in error messages has become. It's _exactly_ the time the user shouldn't get the impression that you're taking a haha oh well attitude.
Humor is subjective, but I've always found Craig's performative sense of humor to be more annoying than anything else. The smug brainiac shtick is so old at this point that the dude practically steps on Steve Ballmer's feet.
> This label is the target of a goto from outside of the block containing this label AND this block has an automatic variable with an initializer
I get this issue a lot on modern compilers, when trying to write switch/macro-based coroutines (https://www.chiark.greenend.org.uk/~sgtatham/coroutines.html) in C++. Does anyone have a workaround? (I hope it doesn't involve C++20 coroutines... I still don't understand them.)
If your variable is a class without a trivial default constructor, then you're out of luck. Otherwise, the workaround is to not initialize the variable, and to instead assign to it immediately after declaring it (e.g., replace "int x = 42;" with "int x; x = 42;").
I remember when I compiled a "hello world" with some new OO framework, possibly MacApp, and it was something like 400K, to do nothing, and I thought that was outrageous and obviously much too bloated for real software.
[1] https://en.m.wikipedia.org/wiki/Green_Hills_Software
[2] https://en.m.wikipedia.org/wiki/Macintosh_Programmer%27s_Wor...
[3] https://web.archive.org/web/20140528005901/http://lists.appl...