feat: improve formatting of error messages from C

[maelle]

Fix #2202

One thing we might consider adding, is a period at the end of the message if it's not already there.

[szhorvat]

One thing we might consider adding, is a period at the end of the message if it's not already there.

I've added punctuation at the end of most messages in the C core now, but a few are still left out. This is a nice-to-have aesthetic feature, it's not an absolute must unless you want to append extra text at the end of the message. We do correct for missing punctuation in both the Python and Mathematica interfaces.

[szhorvat]

Suggested change

	/* Strip vendor/cigraph/src/ prefix from file path */
	static inline const char* simplify_file_path(const char* file) {
	const char* prefix = "vendor/cigraph/src/";
	size_t prefix_len = strlen(prefix);
	if (strncmp(file, prefix, prefix_len) == 0) {
	return file + prefix_len;
	}
	return file;
	}
	/* Strip vendor/cigraph/src/ prefix from file path. This prefix depends on the
	* build procedure, namely on the directory that the compiler is invoked from. */
	static inline const char* simplify_file_path(const char *file) {
	const char prefix[] = "vendor/cigraph/src/";
	const size_t prefix_len = sizeof(prefix) / sizeof(prefix[0]);
	if (strncmp(file, prefix, prefix_len) == 0) {
	return file + prefix_len;
	}
	return file;
	}

You can get the length statically. It's good to have a comment about where this prefix comes from because it will change if changes are made to the build procedure. The reason it even appears (compared to the standard use of the C core) is the tweaked build procedure.

[szhorvat]

One thing we might consider adding, is a period at the end of the message if it's not already there.

Isn't this already there with maybe_add_puctuation?

LGTM, but I didn't test.