W3C home > Mailing lists > Public > public-iri@w3.org > May 2010

Re: [bidi] Special ordering for BIDI URLs

From: Ted Hardie <ted.ietf@gmail.com>
Date: Fri, 28 May 2010 11:24:36 -0700
Message-ID: <AANLkTinF8F1osM9Ow0glRpaeUDIPHSkN5T3iB3O_Z1M8@mail.gmail.com>
To: "Phillips, Addison" <addison@lab126.com>
Cc: (wrong string) ‚˜• <mark@macchiato.com>, Shawn Steele <Shawn.Steele@microsoft.com>, Adil Allawi <adil@diwan.com>, "public-iri@w3.org" <public-iri@w3.org>, "bidi@unicode.org" <bidi@unicode.org>, Murray Sargent <murrays@exchange.microsoft.com>, "aharon@google.com" <aharon@google.com>, Nasser Kettani <Nasser.Kettani@microsoft.com>
On Fri, May 28, 2010 at 10:37 AM, Phillips, Addison <addison@lab126.com> wrote:

>
> Hi Ted,
>
> I'm not sure I'm understanding correctly, but I think you're asking, what if you have SHOP45 (where "SHOP" consists of strongly RTL characters)?

Yes, that's the basic question.

>
> The answer here, I think, is that the Unicode Bidirectional Algorithm does the right/expected thing here. You want the RTL characters presented right-to-left while any LTR characters are present left-to-right. A given token might have a very different presentation vs. logical order, but this is expected and not potentially harmful because you cannot spoof within the token (it is token reordering that's a problem).
>
> That is:
>
> † SHOP45.mycompany.com presents as: 45POHS.mycompany.com
>

Okay, so the registrant in RTL context registers 45POHS in the zone,
and the BiDi
algorithm when it sees this in http://45POHS.LTRexample/
does nothing as nothing is marked as requiring change (NB: the
registration would really be xn--ACE)?
But when it sees this elsewhere in the uri, it relies on a context
marker, so that the ui writes http://another.example/<rtlcontext
marker>45SHOP</rtlcontext marker>
resulting in a display http://another.example/45PHOS?
And we should not present the algorithm something like this:
<rtlcontext marker>http://45POHS.example/45SHOP</rtlcontext marker>
since that would require it special case the domain to avoid
double-reversing the registered name?

Just to confirm, this seems to require someone to have to "manually"
shift the scheme to the correct
side of the authority section, since the rtlcontext markers are split
by the authority section.  So, it
would end up written:  <rtlcontext marker>45SHOP</rtlcontext
marker>/example.45POS//:http
and displayed as 45POHS/example.45POHS//:http ?

That seems pretty painful without an additional marker like
<rtlcontext IRI><rtlcontext marker>45SHOP</rtlcontext
marker>/example.45POS//:http
</rtl context IRI>

So that the IRI could be passed to LTR contexts.

I may, of course, be missing the point here, but I truly appreciate
the time folks
have taken to help bring this out.  I also should not that the xml style markers
above are not meant to indicate that I expect that style to be used--just easy
for me to type.

regards,

Ted Hardie

> And:
>
> †FIRST123SECOND.example.il presents as: †DNOCES123TSRIF.example.il
>
> We do NOT want to reverse the whole string (it isn't 'ptth', it's 'http'). We need to present "runs" in their contextual order but reverse the layout of the runs themselves (or not, as some have contended) in the resultant URI.
>
> Addison
>
> Addison Phillips
> Globalization Architect (Lab126)
> Chair (W3C I18N, IETF IRI WGs)
>
> Internationalization is not a feature.
> It is an architecture.
>
>
>
>
Received on Friday, 28 May 2010 18:25:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 30 April 2012 19:51:57 GMT