* WGs marked with an * asterisk has had at least one new draft made available during the last 5 days

Ticket #25 (assigned defect)

Opened 4 years ago

Last modified 23 months ago

Adapt rules for bidi components to those in IDNA2008 (RFC 5893)

Reported by: duerst@it.aoyama.ac.jp Owned by: adil@diwan.com
Priority: major Milestone:
Component: bidi-guidelines Version:
Severity: - Keywords: bidi
Cc:

Description

This originally was issue http://www.w3.org/International/iri-edit/#bidi-combining-108 in the old issues list. That issue only talked about combining characters at the end of a bidi component. Starting with this case, the IDNAbis work carefully reevaluated the bidi-related constraints on a 'component' (label in the DNS case), resulting is some additional changes. The IRI spec should take up these changes after checking that they work not only with the 'dot' used as a separator in domain names, but also with the various separators used in IRIs.

Change History

comment:1 Changed 3 years ago by stpeter@stpeter.im

  • Component changed from 3987bis to bidi-guidelines

comment:2 Changed 2 years ago by adil@diwan.com

  • Keywords bidi added
  • Owner set to adil@diwan.com
  • Status changed from new to assigned

comment:3 Changed 2 years ago by adil@diwan.com

Starting with this case, the IDNAbis work carefully reevaluated the bidi-related constraints on a 'component' (label in the DNS case), resulting is some additional changes

Is there a link for these changes that I can check?

comment:4 Changed 2 years ago by duerst@it.aoyama.ac.jp

For IDNA and bidi, please see http://tools.ietf.org/html/rfc5893. The sections to look for in particular may be http://tools.ietf.org/html/rfc5893#section-4 and http://tools.ietf.org/html/rfc5893#section-7, but the whole document is worth reading for you (and anybody else interested) anyway.

Please note that we can NOT just use the rules in RFC 5893 directly, because these rules have not been checked against characters such as '/', ':', '?', '#', and so on that appear as component separators in IRIs but not in IDNs. Harald Alvestrand has sent me the script that he used to check this (for '.' as a component separator only), but I have yet to try to adapt it to our needs (it's in Perl, and so the first thing I might do is to change it to Ruby :-(). I'll send you (Adil) a copy privately. If somebody else is still fluent in Perl (I used to be) and wants to give it a try, please contact me.

comment:5 Changed 2 years ago by duerst@it.aoyama.ac.jp

  • Summary changed from Adapt rules for bidi components to those in IDNAbis to Adapt rules for bidi components to those in IDNA2008 (RFC 5893)

comment:6 Changed 23 months ago by duerst@it.aoyama.ac.jp

Splitting out http://trac.tools.ietf.org/wg/iri/trac/ticket/132 (Allow non-spacing marks at end of components), because that's straightforward. The remaining differences, as far as I understand them, are: 1) numbers at the end of labels/components; 2) the way the rule is worded (RFC 5893 uses bidi categories, whereas we currently use just descriptive test).

Note: See TracTickets for help on using tickets.