Re: [PATCH] ldap.el: use auth-source if passwd is not defined in ldap-host-parameters-list

[Ted Zlatanov]

On Thu, 10 Mar 2011 23:05:25 +0200 Leonidas Tsampros <address@hidden> wrote: 

LT> 1) authinfo machine ip-address login username password secret-key imap

LT> and (setq ldap-host-parameters-alist
LT>       (quote (("ip-address" base "dc=domain,dc=com"
LT>                binddn "domain\\username" "auth-source "yes"))))

LT> Logging in to both the LDAP and the IMAP server works as expected (on
LT> the assumption that they are the same hosts, and this is my personal
LT> setup, so I'm pretty happy.

Good.

LT> 2) authinfo machine ip-address login username password secret-key imap
LT>    authinfo machine ip-address binddn domain\username password secret-key 
port ldap

LT> and (setq ldap-host-parameters-alist
LT>       (quote (("ip-address" base "dc=domain,dc=com"
LT>                auth-source "yes"))))

LT> The above is the more general case of having a separate LDAP
LT> server. This case doesn't work, since auth-source-search returns the
LT> first entry and so (plist-get asfound :binddin) returns nothing.

Yeah, we don't require the "ldap" port...  I think it's OK, the user can
reorder them since IMAP does require the port.

LT> In order to work around this case, I thought of the following:

LT> 3) authinfo machine ip-address login username password secret-key imap
LT>    authinfo machine ip-address:389 binddn domain\username password 
secret-key

LT> and (setq ldap-host-parameters-alist
LT>       (quote (("ip-address:389" base "dc=domain,dc=com"
LT>                auth-source "yes"))))

LT> which should work as expected but I think it breaks eudc-ldap. The
LT> module seems trying to match only the 'ip-address' part of the
LT> ldap-host-parameters-alist entry and as such a check to see if a base dn
LT> is defined fails. However I want to double check this again.

I think this has to work so users can specify other LDAP ports.  So we
need to fix eudc-ldap then.  Can you look at what's needed?  I don't use
that package so I'm hardly the one to fix this issue.

Let me know if you agree with the above and I'll push the patch tonight.

Thanks
Ted