On 07/11/06, Martynas Venckus <[EMAIL PROTECTED]> wrote:
> I tried both package and compiled version. It always fails. > I haven't tried your version, I will try it today on amd64. > I'll include kdump log.Thank you for reporting the problem! I'll explain it a bit. The client fails when checking: if (!silc_utf8_valid(client->realname, strlen(client->realname))) { <...>; return FALSE; } silc_utf8_valid returns: return silc_utf8_decode(utf8, utf8_len, 0, NULL, 0) != 0; Snipet from silc_utf8_decode: if (!utf8 || !utf8_len) return 0; Unfortunately, it doesn't ensure that client->realname is filled, when validating. That leads client to just not work by default in the circumstance when real_name is left empty (e.g. pw_gecos on the password database). Below is the new version of silc-client-1.0.3 with the simple patch included to fix the issue. Contacted Pekka Riikonen and hope this will get fixed in the upcomming 1.0.4 (as it affects all platforms). UPDATE: silc-client-1.0.3p1 Tarball: http://www.altroot.org/silc-client.tar.gz Precompiled i386 package: http://www.altroot.org/silc-client-1.0.3p1.tgz Irssi-silc hasn't changed. But just reminding: Tarball: http://www.altroot.org/irssi-silc.tar.gz Precompiled i386 package: http://www.altroot.org/irssi-silc-1.0.3p0.tgz The new patch: http://www.altroot.org/silc-client-irssi-silc.patch
I tested the presented patch for silc-client, indeed, before that it was dying silently when the GECKOS field was empty, works now.
-- Martynas Venckus
-- viq
