Mappings

Shawn Steele Shawn.Steele at microsoft.com
Thu Jul 16 23:40:09 CEST 2009


I would interpret the registration-side "no-mapping" requirement as "The user enters a name, which is then mapped.  The registrar then shows the results for the mapped name to the user, so that they see what they're registering"

In other words, the user ends up seeing exactly what they register.  I don't think that precludes a pre-registration step that makes it easier for the user to enter or request a certain name.  Any such mapping MUST be the same as the "UI" mapping on the lookup end.

Of course, I'm not a registrar, so maybe that isn't best.

-Shawn

-----Original Message-----
From: idna-update-bounces at alvestrand.no [mailto:idna-update-bounces at alvestrand.no] On Behalf Of Chris Wright
Sent: Thursday, July 16,  2009 0:11
To: idna-update at alvestrand.no
Subject: Mappings

WRT mappings, I am strongly in favour of a mappings process being made part of the protocol - I can't think of any reason not to have it and can see MANY problems we faced during implementation that it would help.

When dealing with registrars and the EPP protocol accepting registrations in NFC format with only pvalid, contexto/j etc characters is relatively easy to implement. However, when registrars are using the web interface to our registry, for example, and they come to the search page for instance, at this point it is not always possible for the user to "type" or input in NFC form with only allowed characters. We need to act as an "application" at this point and apply some pre-processing to the inputs. This is very similar if not identical to what a registrar will have to do when they have users hit their websites to register names. I am concerned that without mappings registrars (or different applications) will do different things, and be inconsistent between one another with what they do with the user input they get from potential registrants.

In the context of applications, applications could do things differently, or if it happens further down, at the API level, different OS APIs may do different things (and of course applications can always bypass the OS APIs and lookup names directly).

So when registering names, if there are no mandatory mappings, registrars will need to "guess" what each browser, email client, OS etc will do when it is given the input that the user has typed on the domain registration page and then try to convert that input to the most appropriate valid domain name. If the registrar doesn't do this and expects the name to be supplied in a valid form (as the registry does), then the registrant will need to do it. Regardless, at some point, someone needs to do some transformation and if there are no mandatory mappings, each entity doing these transformations will need to try to guess what each other entity will do and I just don't see this working.

Mappings should be specified such that they MUST be applied on LOOKUP (and they SHOULD be applied at registration but not by the registry).

All those that are concerned about storage of IDNs in URIs, email addresses, address books and other databases etc should not be worried, entries can be stored as entered by the user, once that entry is retrieved from "storage" and then 'looked up' the mapping should be applied at that point before the query is converted to puny code and sent to the DNS.

I am strongly in favour of ALL the mappings being 'MUST' I simply cannot fathom how we intend this to feasibly work without this requirement. How will users determine which name to register? and I don't understand the arguments people are making for these only need to be SHOULDs or even worse MAYs, am I missing something?

Finally I support the suggested reordering of the steps in the mapping document either casefold/lowercase, width map then NFC or width map, casefold/lowercase then NFC, the closer to IDN2003 the better however if valid reason exist to do something different then so be it, though I do believe that NFC should be the last step. I don't as yet understand technically what the differences will be between lower case/case fold enough to have an opinion on which should be used - I am still looking into this

Thanks

Chris Wright
Chief Technology Officer
AusRegistry Pty Ltd


_______________________________________________
Idna-update mailing list
Idna-update at alvestrand.no
http://www.alvestrand.no/mailman/listinfo/idna-update



More information about the Idna-update mailing list