Re: Too much 'E'? was Re: Lists and EPP

On 8 jun 2011, at 15.47, Edward Lewis wrote:

> At 13:32 +0200 6/8/11, Patrik Fältström wrote:
> 
>> People have, we see now, used the "e" part of epp a bit too much.
> 
> Can you quantify this?  (Asking out of curiosity.)  As in, has there been a failure in the protocol?  Has the goal of interoperability failed?

Yes. I claim that have failed from some point of view. The protocol is interoperable, but what you transfer inside the protocol is different.

Just look at Net::DRI in Perl and you see that every TLD have their own extension. They have their own rules for for example how a domain name is to be written (do they accept [a-z] only, or [a-zA-Z]?). Etc...

I think personally it is because all development of the protocol, and the extensions, are made by the registries, and not enough input from registrars which are the ones that today have to pay the burden of what I call non-interoperability.

Hey, some registries do not even release their specification unless you pay to become a registrar!

Yes, different business models exists for the registries, but I do not see such diversity is needed as exists today in for example the epp specs for the various TLDs.

> Between 2009 and 2010 I poked into EPP 2.0 and found that everyone was happy with EPP 1.0 and the plethora of extensions, especially with Patrick Mevzek (I believe) has implemented the bulk of them in one distribution.

Who are everyone? How many registrars did you talk with that actually have implemented epp from ground up that have connections with say 5-10 different registries where a minority are gTLDs?

>> And I dare to say that as I as Area Director when epp was invented could
>> have pushed back more, but I did not. So I blame myself.
> 
> As one of the co-chairs, I'm willing to pass the blame on. ;)

Ha ha ha!

> At 13:55 +0200 6/8/11, Patrik Fältström wrote:
> 
>> So yes, we needed extensions, I just think registries could have done better
>> if they coordinated a bit more the standardization of the extensions, instead
>> of creating one epp extension per registry/domain name combination.
>> 
>> To conclude: I agree that epp is probably not the most efficient protocol to
>> do sync:ing, but I rather see one syncing extension than "one per registry".
>> That was my (only) point.
> 
> I continued in the thread and found this.
> 
> I thought the same as you expressed when I did some organization for a non-BoF at the Anaheim IETF.  Architecturally speaking, I had some misgivings about the path EPP had taken.  From March 2010 until May 2010 some efforts ramped up as a result of the IETF-related meeting, including a new mailing list started called regops.  In early May, in a CENTR Tech meeting some preliminary results were presented - mostly the estimate of what it would take to "fix" EPP. This presentation was a follow up to a discussion during the October 2009 CENTR Tech meeting that kind of kicked off this short study.
> 
> The discussion at the May 2009 CENTR Tech meeting came down to "okay, well, EPP 1.0 is sufficient."

CENTR is biased to registries.

> I took this to mean - in the operations world, EPP is just fine.  Operators are satisfied that there is code to do all the versions needed.  To an operator having two different "redemption grace period" extensions, one for COM and one for BIZ, is not a problem.

Yes, it IS a problem.

To have one extension for every TLD, that create a nXm matrix of number of extensions times number of TLDs.

> Maybe as an architect it sounds ugly - but then who are protocols built for, the architects or the operators?
> 
> After that meeting I abandoned any upgrade/update to EPP and learned to live with the way it is. Code rules, specs aren't as important.

And I think you made the correct decision. Or you would go mad! ;-)

   Patrik

_______________________________________________
provreg mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/provreg