Read on Twitter
Well hello there, I'd like to share some code I wrote in, oh gosh, 2008. I don't actually have the code any more. But here's a picture I drew of the architecture
I did find somebody made a copy on github, and I think it's missing a bunch of stuff. but it has my favourite parts!
Here's my implementation of C89's malloc() — as you can see it returns NULL based on a probability:
Here's my implementation of C89's malloc() — as you can see it returns NULL based on a probability:
The probability there is seedable and deterministic, and doesn't affect rand(). Which means I had to keep my own prng internally, and provide rand() separately, too.
So here's my C99 rand()! It sometimes returns 7.
So here's my C99 rand()! It sometimes returns 7.
The C99 malloc is more complicated! Because there's an extra clause about IDB when you request an allocation of 0 bytes.
If you ever need a valid pointer value to return, but you'd like to invoke UB whenever somebody dereferences it, you can do it like this, and return great_nothing + 1.
It looks like my gettimeofday() returns (again, seedable and deterministic) a random time, because I couldn't find a clause that required it to be monotonic.
If you ever have the pleasure of writing a specification, please imagine people like me are sitting on your shoulder
If you ever have the pleasure of writing a specification, please imagine people like me are sitting on your shoulder
Anyway this person's copy of my code taken at some arbitrary revision is here, if you'd like to see! I quite understand if you'd rather not. hehe
https://github.com/lispmeister/libgreat
https://github.com/lispmeister/libgreat
Oh what do you know, he's on twitter.
Hello @lispmeister!! Thirteen years ago you did something that's made my day a little nicer today. Thank you very much!
Hello @lispmeister!! Thirteen years ago you did something that's made my day a little nicer today. Thank you very much!
ps. This is how to comment code when you're writing to a spec.
