On Wed, Mar 07, 2012 at 04:48:45PM +0100, Thijs Kinkhorst wrote:
> I'm not so sure about that. In the deployments I've seen there's usually
> the givenName, sn (common name) pair, something with initials; and the cn
> (common name, the full name or usual name someone goes by).
> As we're Debian I've checked the core schema as shipped with Debian
> openldap and this defines those attributes and as far as I can see doesn't
> create a concept of "middle name".
> I would find it reasonable for the Debian LDAP to only carry the cn as
> this accomodates the possible uses Debian has for this data, it
> accomodates people with one-word names and in my eyes yields just what you
> want: a string representation of the common name someone goes by. If more
> distinction of the last nameis required for some reason, givenName + sn
> will allow that. In any case I don't yet see why there's a need to add a
> middleName as a field.
I've mentioned your email on DSA's IRC channel, and here is a quick feedback:
@sgran> I'd be happy to change where we stick things if we had a better idea
who the consumers were and what impact this change would have
@weasel> also, somebody who isn't us needs to convert ldap
@weasel> and needs to provide patches for our tools
@weasel> also, if this means we can get rid of the UTF-8 gecos field, that
would be really great.
@weasel> gecos is an ascii field defined in the relevant standard schemas
@weasel> and we overwrite it, which breaks stuff every now and then
So I reckon that there is space for improvements, as long as it has good
practical outcomes and as long as one would like to help with the conversion
I'd like to take myself off as middleman in this discussion; of course I'll be
happy to adjust the nm.d.o database and interface to follow any changes that
happen in LDAP, but I can't do much more than that, so I'm passing the ball to