getopt_long returns an int, not a char, and this matters a lot

[nwf-msr]

httpdirfs/src/main.c

Line 172 in d1a10d4

char c;

contains a tragically common and subtle error; it should be int, not char, so that the comparison to -1 in

httpdirfs/src/main.c

Lines 204 to 206 in d1a10d4

	while ((c =
	getopt_long(argc, argv, short_opts, long_opts,
	&long_index)) != -1) {

works portably. As written, it is broken on Arm for example, and the compiler will instead generate an infinite loop; see, for example, https://poudriere.cheribsd.org/hosts/freebsd-arm64/data/latest-per-pkg/fusefs-httpdirfs/1.2.3/cheribsd-aarch64-main.log, which says:

src/main.c:206:39: warning: result of comparison of constant -1 with expression of type 'char' is always true [-Wtautological-constant-out-of-range-compare]
                        &long_index)) != -1) {

Unfortunately, this small error means that we always take the error path at

httpdirfs/src/main.c

Lines 76 to 81 in d1a10d4

	if (parse_arg_list(all_argc, all_argv, &fuse_argv, &fuse_argc)) {
	/*
	* The user basically didn't supply enough arguments, if we reach here
	* The point is to print some error messages
	*/
	goto fuse_start;

and, crucially, skip over the rest of the setup, both of the mountpoint and the network. For maximum confusion, though, because we haven't actually done anything wrong (because we haven't done anything at all, in some sense), the process ends up parked reading from /dev/fuse as it should, but without the mountpoint established, so it's listening for a message that's never coming.