change struct initialization makes more explicit which fields we're setting. (and kill an extra empty line)

use memset(3) rather than bzero(3) There's no difference, but bzero(3) says STANDARDS The bzero() function conforms to the X/Open System Interfaces option of the IEEE Std 1003.1-2004 (“POSIX.1”) specification. It was removed from the standard in IEEE Std 1003.1-2008 (“POSIX.1”), which recommends using memset(3) instead. so here we are.


make sure l is always initialized I can't think of cases where we reach serialize_iri and path is NULL, but let's keep the safe side and initialize l. gcc 8 found this, clang didn't.

fix IRI-parsing bug Some particularly crafted IRIs can cause a denial of service (DOS). IRIs which have a trailing `..' segment and resolve to a valid IRI (i.e. a .. that's not escaping the root directory) will make the server process loop forever. This is """just""" an DOS vulnerability, it doesn't expose anything sensitive or give an attacker anything else.

fix various compilation errors Include gmid.h as first header in every file, as it then includes config.h (that defines _GNU_SOURCE for instance). Fix also a warning about unsigned vs signed const char pointers in openssl.

[cgi] split the query in words if needed and add them to the argv

[iri] accept also : and @ again, to be RFC3986 compliant.

don't %-decode the query

bring the CGI implementation in par with GLV-1.12556

ensure isn't NULL

accept a wider range of UNICODE codepoints while parsing hostnames

legibility: use p[n] instead of (*(p + n))

trim_req_iri: set error string

trim initial forward slashes this parse gemini:// into an IRI whose path is "foo". I'm not 100% this is standard-compliant but: 1. it seems a logical consequence of the URI/IRI cleaning algo (where we drop sequential slashes) 2. practically speaking serving file a sequence of forward slashes doesn't really make sense, even in the case of CGI scripts