-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

This will make the lives of downstream distributors easier. Since Monkeysphere aims to be a combination of libraries, programs, scripts, and documentation, it makes sense that different parts may be written by different people and subject to different copyright and license terms. Thus a good scheme for keeping track of this information is very much desired.
-----BEGIN PGP SIGNATURE-----

iHUEARYKAB0WIQSiPzylvTnZ6xisfzWz9N0oYfTNugUCZNnC0wAKCRCz9N0oYfTNuj6IAP9Ni6r6GyBnLRyBKlVFdLiJLd6kOCOVHh0Ecz19CBwqQQD7BnNVykYa2lqjkk0e3ITS1h3tHpM4LsRahm44U9qhpAA==9UI0
-----END PGP SIGNATURE-----

One of the most general needs of a build system is to test that particular functionality is supported, and which the build system provides no other way to probe for. As such, I would appreciate if the run() method for compiler objects could be implemented.

Howdy,

Some other small compilers (TinyCC and maybe ChibiCC) support automatic allocation of flexible array members upon initialization, and it would be really nice if cproc would support this as well, or at least error upon trying to use this construct.

On my x86_64 Debian Bookworm GNU/Linux system, the following program compiles successfully but segfaults at runtime:

int main(void) {
	struct {
		int i;
		char s[];
	} w = { .s = "Hello, world" };
}

With TinyCC, this does the common sense thing of allocating space for the string on the stack. GCC and Clang only support this construct if the struct is static. In this case though, we get a nifty assertion failure:

cproc-qbe: qbe.c:1449: emitdata: Assertion `offset <= d->type->size' failed.
cproc: compile: process signaled: Aborted