mild kestrel
What do y'all think about the following syntax for requiring that a pointer argument is non-null?
(gcc/clang will issue a warning if they can detect that you're passing NULL)
void thing_do_stuff(thing_t thing[static 1]);
Would you consider this readable?
Would you easily understand what's happening if you encountered this in a library?
Would a macro make it better or worse?
#define MYLIB_NONNULL static 1
void thing_do_stuff(thing_t thing[MYLIB_NONNULL]);
jaunty aurora
a) i don't really think this is readable without some kind of macro
b) msvc won't compile this, which is entirely their fault but a thing to consider nonetheless
macro options i would put into the ring besides what you put: c #define MYLIB_NONNULL [static 1] void thing_do_stuff(thing_t thing MYLIB_NONNULL); the array syntax itself is unintuitive to me. but this reads a bit funny to me, so you could convert to a function like macro: ```c
#define MYLIB_NONNULL(param) param[static 1]
void thing_do_stuff(MYLIB_NONNULL(thing_t thing));
a problem i have with both of those is now the indirection is hidden now that i look at it properly. so idk maybe the array syntax with inner macro is best
mild kestrel
i don't really think this is readable without some kind of macro
that's what I was thinking. I'm guessing many people either don't know about this use ofstaticin C, or wouldn't immediately make the connection between "at least one member" and "cannot be null".
a problem i have with both of those is now the indirection is hidden now that i look at it properly
yeah, unfortunately, there's no good way to indicate that the parameter is a pointer besides using the "array" syntax or just spelling it out:MYLIB_NONNULL_PTR(param)
msvc won't compile this
wait--what?
checks godbold
oh my god you're right
(so much for "supporting all required features"...)
well, thanks for the input/ideas (and pointing out my oversight!)
just one more reason to hate microsoft, I suppose...