>>>>> "William" == William M Perry <wmperry(a)aventail.com&gt; writes: William> Michael Harnois <mharnois(a)willinet.net&gt; writes: >> On 07 Jun 1999 07:38:04 -0500, wmperry(a)aventail.com (William M. Perry) said: >> >> > ("LDAP error", >> > build_string (ldap_err2string (ldap_get_lderrno (ld, NULL, NULL)))); >> >> If I read my files correctly the function that is here called >> ldap_get_lderrno is in OpenLDAP 2 called ldap_result2error. I also >> noticed in other examples in the OpenLDAP source which used the function >> that the third argument is 0 rather than NULL. William> well, ldap_result2error only gets you some of the cases where an error William> occurs. What happens if you want to see why ldap_init() failed? Bill, it seems we've been working on the same thing at the same time, we probably should synchronize a little better in order to avoid wasting our efforts. In the case ldap_init() fails errno should be checked. This is what we do already for ldap_open(). My patch is a little more extensive than yours and handles the errors in compliance with the current IETF draft-ietf-ldapext-ldap-c-api-02. Furthermore if I read it well, yours should fail if you use libraries without liblber.

However, from you patch I saw things that would need to be changed in mine: - remove the check for libldap10 as you appropriately mentioned - use AC_CHECK_FUNCS instead of individual AC_CHECK_FUNCs - probably test and use ldap_init() instead of ldap_open() since the draft deprecates this function (I'd rather wait for an RFC here however) What I propose is for the CVS maintainers to hold on until I come up with something that fixes all this. Probably still today.