It's been quite a challenge to maintain both radiusNTPassword and sambaNTPassword in a user's LDAP object, especially when radiusNTPassword is just sambaNTPassword, prepended with '0x'. If nothing else, that's redundant. So, what about this patch? It just takes your sambaNTPassword and prepends the necessary '0x' before doing anything with it. Does that make sense? Anything wrong with doing this? Obviouly this patch is teeny and not very profound... For what reason is this avoided? rlm_ldap expects the value to be in hex. OK -- so just immediately assume it's in hex and prepend the '0x'. What about making it a configuration parameter? I'm sure others would like to be able to skip the whole redundant *NTPassword thing, too. Comments encouraged. Thanks!! Stefan # cat /tmp/0x.patch diff -urN freeradius-1.1.7/src/modules/rlm_ldap/rlm_ldap.c freeradius-1.1.7-0x/src/modules/rlm_ldap/rlm_ldap.c --- freeradius-1.1.7/src/modules/rlm_ldap/rlm_ldap.c 2007-08-08 22:57:43.000000000 -0500 +++ freeradius-1.1.7-0x/src/modules/rlm_ldap/rlm_ldap.c 2007-08-08 22:37:56.000000000 -0500 @@ -2498,6 +2498,7 @@ int vals_count; int vals_idx; char *ptr; +char hex[35]; char *value; TLDAP_RADIUS *element; LRAD_TOKEN token, operator; @@ -2517,6 +2518,7 @@ */ if ((vals = ldap_get_values(ld,entry,element->attr)) == NULL) continue; +DEBUG("!!! %s !!!", element->attr); /* * Check whether this is a one-to-one-mapped ldap * attribute or a generic attribute and set flag @@ -2535,6 +2537,9 @@ for (vals_idx = 0; vals_idx < vals_count; vals_idx++) { value = vals[vals_idx]; +sprintf(hex, "0x%s", value); +value = hex; +DEBUG("!!! %s !!!", value); if (is_generic_attribute) { /* * This is a generic attribute.