[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: ACM permission



Helmut,

It works with DirX!

I am talking about authentication for an application, not the Directory
itself.  To clarify: If I have the entry in another DSA, but access some
other DSA, a Compare operation will be propagated as a chained compare to
the DSA that holds the entry I am trying to validate against.

It may not be the best solution from a security perspective, but the Model
allows for it.

Cheers,                  ....Erik.


Erik Skovgaard
Siemens Meta-Directory Solutions
Phone: +1 604-204-0750
Fax:   +1 604-204-0760

-----Original Message-----
From: Volpers, Helmut [mailto:helmut.volpers@icn.siemens.de]
Sent: Thursday, July 12, 2001 07:01
To: 'Skovgaard, Erik'; 'Kurt D. Zeilenga'
Cc: 'Mark Davidson'; ietf-ldapext@netscape.com
Subject: RE: ACM permission


Hi Erik,

> -----Original Message-----
> From: Skovgaard, Erik [mailto:Erik.Skovgaard@icn.siemens.com]
> Sent: Donnerstag, 12. Juli 2001 15:37
> To: Volpers, Helmut; Skovgaard, Erik; 'Kurt D. Zeilenga'
> Cc: 'Mark Davidson'; ietf-ldapext@netscape.com
> Subject: RE: ACM permission
> 
> 
> Helmut,
> 
> I am working with at least two products that use the compare 
> operation to
> validate a password.  Agreed, that is not the best way, but 
> the point here
> is that the Directory was certainly intended to support this 
> authentication
> method and we should not preclude it.

I agree that application use this to validate a password, but I don't think 
that any LDAPServer handle this as an authentication.
> 
> I am not aware of any chained Bind operation, but my latest 
> X.518 document
> date back to 1993.  Are you telling me that the operation has 
> been added in
> later versions?

No. But if you want to use chaining for authentication (what I don't like)
why not doing it with a bind instead of a compare. 

Helmut
> 
> Cheers,                   ....Erik.
> 
> Erik Skovgaard
> Siemens Meta-Directory Solutions
> Phone: +1 604-204-0750
> Fax:   +1 604-204-0760
> 
> -----Original Message-----
> From: Volpers, Helmut [mailto:helmut.volpers@icn.siemens.de]
> Sent: Thursday, July 12, 2001 02:13
> To: 'Skovgaard, Erik'; 'Kurt D. Zeilenga'
> Cc: 'Mark Davidson'; ietf-ldapext@netscape.com
> Subject: RE: ACM permission
> 
> 
> Hi Erik,
> 
> It's not the normal way to use a compare operation on the password for
> authentication.
> Why not use the bind and you have no problems with AccessControl.
> 
> I am not sure whether you want to make a chained bind, but if 
> you do it it
> is
> a chained bind and not a compare operation on the userPassword.
> 
> Helmut 
> 
> > -----Original Message-----
> > From: Skovgaard, Erik [mailto:Erik.Skovgaard@icn.siemens.com]
> > Sent: Dienstag, 10. Juli 2001 23:00
> > To: 'Kurt D. Zeilenga'
> > Cc: 'Mark Davidson'; ietf-ldapext@netscape.com
> > Subject: RE: ACM permission
> > 
> > 
> > Kurt,
> > 
> > I have applications that use the compare operation on the 
> > userPassword for
> > authentication.
> > 
> > BTW, a BIND may result in a compare operation if you use 
> > chaining on the
> > back end of the server.  Has anyone considered that?
> > 
> > Cheers,                  ....Erik.
> > 
> > Erik Skovgaard
> > Siemens Meta-Directory Solutions
> > Phone: +1 604-204-0750
> > Fax:   +1 604-204-0760
> > 
> > -----Original Message-----
> > From: Kurt D. Zeilenga [mailto:Kurt@OpenLDAP.org]
> > Sent: Monday, July 09, 2001 13:17
> > To: Skovgaard, Erik
> > Cc: 'Mark Davidson'; ietf-ldapext@netscape.com
> > Subject: RE: ACM permission
> > 
> > 
> > At 09:25 AM 7/9/2001, Skovgaard, Erik wrote:
> > >That would be a problem.  A lot of us still use the 
> userPassword for
> > >authentication.  It must be possible to protect the password 
> > (including
> > >performing filter matching) yet be able to use the compare 
> > operation on the
> > >attribute.
> > 
> > I'm not sure how permissions for compare relate to authentication.
> > The only operation which performs LDAP authentication is the
> > bind and its not controlled, per the I-D, by any permissions.
> > 
> > This said, I support having separate "assert" (compare/search
> > filter) permissions from read permissions as it is often useful
> > to allow one to assert a value but not allow them to read all
> > values.  The example (which I believe someone else gave) is
> > that there may a group where one is allowed to assert that
> > an entity is a member but not allowed to see the member list.
> > 
> > Kurt
> > 
> > 
> 
> 

Attachment: Skovgaard, Erik.vcf
Description: Binary data