IDNAbis spec

Shawn Steele Shawn.Steele at
Thu Nov 5 20:00:31 CET 2009

My main point is that I think consistency is important.  Perhaps rendering labels in LTR order always would work, perhaps rendering in RTL order if there's any RTL characters would work.  The problem is that when some parts are rendered in one direction and others in another direction, then it becomes terribly unclear what the URL is doing.

Martin mentioned that users had to learn what meant, and I think they can adapt to whatever ordering happens, so long as it is reasonable.

I also think this needs a lot of thought, because I'm not sure at all that this is very clear in terms of a full IRI.  I certainly haven't thought about it enough.  In a full URL I think it's less clear because we don't necessarily know what the server's using for delimiters.  What happens with "" ?  If the whole thing's RTL, do you render as fun=flags&shawn=user?html.mypage/com.mydomain//:http ?  What if the "flags=" and "user=" stay LTR?  And then there're all the systems that use completely different syntax to do that type of thing.  Even saying IRIs have to be LTR is hard because individual words obviously can't be LTR if they're RTL words, so in that case, what's being a delimiter and what's part of a word?  Is _ a delimiter or word part?


From: idna-update-bounces at [idna-update-bounces at] on behalf of Slim Amamou [slim at]
Sent: Thursday, November 05, 2009 9:06 AM
To: Martin J. Dürst
Cc: Alireza Saleh; muhtaseb at; public-iri at; idna-update at; Lisa Dusseault; Shawn Steele; Abdulrahman I. ALGhadir
Subject: Re: IDNAbis spec

hi everybody,

On Wed, Nov 4, 2009 at 6:57 PM, Shawn Steele

I think that labels are like a list.  If I have a list (a, b, c, d), then I expect the list to be in order a, b, c, d.  In an RTL context I would reasonably expect the list to be rendered (d, c, b, a).  If the individual values happen to be in different scripts, that's not going to change the fact that I expect the list to have each element progress in an orderly fashion from least significant to most significant.

This is actually not obvious, for example for numbers, the order is left to right and arabic readers don't have any problem with that.

In my opinion, enforcing LTR on IRIs is a better solution than //:http. An it seems that relevant specs (RFC3987 and w3c IRI bidi examples) are on my side.

I raised that issue a while back on IDN wiki. You can see the discussion and the relevant specs here :

Slim Amamou | سليم عمامو
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the Idna-update mailing list