Re: [ldapext] DBIS - new IETF drafts

[Mark R Bannister <[e-mail]>]

On 10/01/2014 00:42, Luke Howard wrote:
> I was reluctant to join this conversation as I haven't had a chance to look at the drafts, but: to anyone wanting to improve on RFC2307 (or indeed, anything), I would say go for it.

Thanks Luke, I went for it already.

> To get traction IMO, it needs to (a) be published in a finished state (unlike 2307bis)

That bit might be hard, even once the reference implementation is 
complete and it's proven how does one get something out of draft into a 
published state?

> , and (b) have a reference implementation that supports every platform.

Working on it.  As much as possible with be Python, so apart from 
NSS/PAM semantics, it will be quite portable.

> Enterprises that are still welded to NIS and need all the corner cases (netgroups, case sensitivity, etc) are also likely to still be using AIX, HP-UX, Solaris 8, etc. (Of course there is always inertia owing to the installed base, but were one to have that attitude about everything, there would be no progress. In the end the market will decide.)

Indeed, and all of the big clients I've worked for still have a sizeable 
Solaris 8 footprint, and those clients are very reluctant if not 
refusing to upgrade that estate.  I haven't seen as much AIX or HP-UX.  
I would rather DBIS supports all of the legacy platforms as well, so 
that everyone can make a clean break from NIS and RFC2307 rather than 
having to support both environments.  This is why I have focused very 
hard on compatibility between the models.

Best regards,
Mark.