Re: Status of LDIF and Changelog?

[Gordon Good <ggood@netscape.com>]

The timing of this thread is good!

The plan I have, as document author, is to move the current LDIF draft forward
as a proposed standard. I apologize for not getting this done sooner, but other
things have gotten in the way. Here are some things to think about:

- Is proposed standard the way to go?  I think so. Having LDIF on standards
track doesn't preclude other formats (like an XML DTD for directory entries)
coming into existence. Instead, it's an acknoweledgement that many
implementations already support LDIF, and will help ensure interoperability
among implementations that choose to support LDIF.

- What about the LDIF extensions draft
(http://search.ietf.org/internet-drafts/draft-andersen-isss-ws-dir-ldifext-00.txt)
that Erik Andersen has written? Erik's draft addresses three main issues: it
allows "ownership" of attribute values to be described in an LDIF file (to
allow merging of data from multiple organizations), it allows for mapping
between attribute names, and is more compact. The first two features enable
LDIF to be used to perform certain metadirectory operations. The last is a
feature which is orthogonal to the other features. In my opinion, these are
interesting and potentially valuable features, but I would not want to see them
become part of the basic LDIF specification. Doing so would unnecessarily
burden implementors who do not require these features.

- The changelog draft, in my opinion, should become an informational RFC. The
LDUP group is not planning to use LDAP-accessible changelogs in its
multi-master replication work.

How does this sound?  Are there any serious objections to these plans?