Return-Path: X-Original-To: apmail-infrastructure-dev-archive@minotaur.apache.org Delivered-To: apmail-infrastructure-dev-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 1F6D06D0F for ; Thu, 9 Jun 2011 10:42:44 +0000 (UTC) Received: (qmail 61340 invoked by uid 500); 9 Jun 2011 10:42:43 -0000 Delivered-To: apmail-infrastructure-dev-archive@apache.org Received: (qmail 61224 invoked by uid 500); 9 Jun 2011 10:42:43 -0000 Mailing-List: contact infrastructure-dev-help@apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: infrastructure-dev@apache.org Delivered-To: mailing list infrastructure-dev@apache.org Delivered-To: moderator for infrastructure-dev@apache.org Received: (qmail 13449 invoked by uid 99); 9 Jun 2011 08:44:20 -0000 MIME-Version: 1.0 In-Reply-To: References: From: =?ISO-8859-1?Q?N=F3ir=EDn_Plunkett?= Date: Thu, 9 Jun 2011 10:44:00 +0200 Message-ID: Subject: Re: LDAP custom-asf schema - family name? To: sebb Cc: infrastructure-dev@apache.org, ASF Site-Dev , Tony Stevenson Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable On Thu, Jun 9, 2011 at 9:35 AM, sebb wrote: > Tony wrote that he is working on the custom-asf schema currently. > > It would be very useful for generating sorted lists of names if LDAP > contained each person's given name and family name - or at least the > family name and the full name. Eeep! I see many unspoken assumptions here that I think we need to be careful abo= ut. Sorting names is *hard*. Hell, even splitting names into "given" and "family" has pitfalls, and it's certainly not always appropriate to sort on "family name". > Is that something that exists already, or is being considered? > > Also, there is a distinction between a person's full name (legal name) > and their name as they wish it to be known publicly. These are not > always the same. The distinction between legal name (which may or may not actually be the same as full name) and public name is, as far as I know, already being accounted for in the planned schema. > For the purposes of generating public output, of course the public > name should always be used. Agreed, of course :-) > This is not always directly related to the full name, so it implies > that the family name should either be public or there need to be > separate versions for the full and public versions. I think you're suggesting that output should be sorted on family name. This is a) assuming that everyone *has* a family name, which isn't always the case; b) inappropriate for several cultures, particularly those where names are patronymic; c) not as straightforward as you think even if everyone does have a family name and none of them are patronyms. For example, what's the correct sort order for some (actual :-) ) friends of mine? Van Riper, Christian van den Bosch, Hugo van der Merwe, Paul Fenwick (Yes, there's more than one correct answer.) I think it's best to stick to "legal name" and "public name", and simply do a string sort on public name. N=F3ir=EDn