Re: (ITS#3791) start_tls while chasing referrals

[ando@sys-net.it]

> rhafer@suse.de wrote:
>> Hi,
>>
>> any news on this? Is it planned to integrate this patch into CVS? To me
>> it
>> look reasonable.
>>
>>
> I was looking at the patch and wondering why we don't just test
> ld->ld_defconn here; it seems the test on ld->ld_sb is unnecessary.
> Didn't have time to investigate further though. Anyway, at present the
> code isn't clear to me. If someone else is comfortable with it, they
> should speak up.

if one calls ldap_init[ialize](), ld_sb is created, but ld_defconn is not;
if one calls ldap_open(), they both get created, and ld_defconn gets ld_sb
as Sockbuf.  Since one can call ldap_start_tls[_s]() right after
ldap_init[ialize](), the test on ld_sb seems reasonable.  I don't quite
see if and why the Sockbuf in ld_defconn can differ from ld_sb in the
first place, so this fix seems useless.  Also, I note this control is
missing from ldap_start_tls(), the asynchronous version of
ldap_start_tls_s().

p.
-- 
Pierangelo Masarati
mailto:pierangelo.masarati@sys-net.it


    SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497