RE: Re: rtli and ltri

Personally, I don't like dir="ltri" or dir="rtli". This is overloading 2 properties on the same attribute. It may offer some advantage to implementers, but the odd cases mentioned by Aharon show that this is not a good design, IMHO.

If one prefers having the isolation expressed as an attribute, then let us create a specific attribute for it, like bdi="yes"|"no", with default values which could be dependent on the type of element affected.

 

Shalom (Regards),  Mati

 

From: Aharon (Vladimir) Lanin [mailto:aharon@google.com] 
Sent: Tuesday, May 08, 2012 5:18 PM
To: Richard Ishida
Cc: public-i18n-core@w3.org
Subject: Re: Re: rtli and ltri

 

I like it too, but if <bdi> is still going to be available, what should <bdi dir="rtl"> mean - the same as <span dir="rtli"> or <span dir="rtl">? The latter makes no sense - bdi is supposed to mean "bidi *isolate*", but the former also seems strange, since the user specifically said  <bdi dir="rtl">, *not* <bdi dir="rtli">.

 

Also, does this mean that you are no longer trying to get markup that bidi-unaware users can be convinced to do without specific reference to bidi that will make their HTML work properly with bidi data?

 

Aharon

On Tue, May 8, 2012 at 4:07 PM, Richard Ishida <ishida@w3.org> wrote:

On 05/04/2012 08:52 AM, Richard Ishida wrote:

       I've been niggled for a while now by the asymmetry of the bidi
       contructs we have in html5, and the more I look at our html5
       recommendations, the more it seems to me that the bdi element is
       really not the most simple way to do isolation when you know
       the direction of text.

       For example

       <p>ltr-text <cite><bdi dir=rtl>RTL-TEXT</bdi></cite><__/p>



       is convoluted and verbose. It also allows the possibility of
       introducing unwanted spaces between the element tags when
       pretty-printing code or other editorial operations.

       It seems to me that, for the situations where we know the
       direction in particular, it would have been easier to simply invent
       two new values for the dir attribute: rtli and ltri.



=====================





On Fri, May 4, 2012 at 6:57 PM, fantasai <fantasai@inkedblade.net

<mailto:fantasai@inkedblade.net>> wrote:

   I like it!

   ~fantasai


=====================



On 06 May 2012 12:43:47 +0300, Aharon (Vladimir) Lanin wrote:


No, we never considered it. Yes, it does have some things going for it.

Keep in mind, though, that when we asked the HTMLWG for isolation in
HTML, we asked for a new attribute. They decided that they liked a new
element better.

Another thing is that this may conflict with an element-based approach.
That is, if dir="ltri" means {unicode-bidi:isolate; direction:ltr}, but
dir="ltr" means  {unicode-bidi:embed; direction:ltr}, it seems strange
that <output dir="ltr"> nevertheless uses unicode-bidi:isolate.

Aharon