Re: ldap_passwd.diff

[Timur I. <Tim...@oi...>]

On Sun, Apr 16, 2006 at 12:46:44PM +0200, Michael Str=F6der wrote:
> Timur Izhbulatov wrote:
> > On Thu, Apr 13, 2006 at 07:47:05PM +0200, Michael Str=F6der wrote:
> >=20
> >>1. As long as returning a server-generated password is not implemented
> >>it does not make sense to make newpw optional and/or accept None as val=
ue.
> >=20
> > Agree. I just blindly followed the RFC.
>=20
> If we can't make newpw an optional key-word argument we also can't make
> user and oldpw to optional key-word arguments.

Yes, now I see. At this moment there is no point in making any of the
arguments optional.

> >>2. Personally I'd like avoid to turn arguments user,oldpw,newpw of
> >>passwd() into optional key-word arguments (and we can't do that for only
> >>user and oldpw, see 1.). I'd rather prefer the application developer to
> >>really know what he's doing. But I'm open to other opinions.
> >=20
> > In this case the application developer won't be able to do some importa=
nt
> > things. For example, changing other users's passwords will be impossibl=
e even if
> > tha application is bound with root DN.
>=20
> The developer could simply pass value None to passwd() for user and oldpw.

Passing None is OK as long as l_ldap_passwd() allows this. So I suggest app=
lying
my changes only to Modules/LDAPObject.c.

Cheers,

--=20
Timur Izhbulatov
OILspace, 26 Leninskaya sloboda, bld. 2, 2nd floor, 115280 Moscow, Russia
P:+7 495 105 7245 + ext.205 F:+7 495 105 7246 E:Tim...@oi...
Building Successful Supply Chains - One Solution At A Time.
www.oilspace.com