W3C home > Mailing lists > Public > public-i18n-bidi@w3.org > January to March 2010

Re: FPWD of Additional Requirements for Bidi in HTML (make <a> <a bdi="yes">)

From: Najib Tounsi <ntounsi@gmail.com>
Date: Thu, 18 Mar 2010 15:58:23 +0000
Message-ID: <4BA24D9F.2010109@emi.ac.ma>
To: "Aharon (Vladimir) Lanin" <aharon@google.com>
CC: public-i18n-bidi@w3.org

I suggest "SCE", because I think the term of self-contained entity is a 
keyword for the introduction of the new html feature. You want to 
isolate a piece of bidi text, because you want it to be displayed 
altogether.
That why I say, somewhat exaggeratedly, SCE is to BDI, what <em> is to 
<i>, i.e. think in term of meaning rather than in term of display.

In connection with this, I would consider a text link as SCE too. i.e. 
making <a ...> <a bdi="yes">.

Indeed, consider the "PIZZA - 3 Reviews" example of §2.1 (displayed as 
"3 - AZZIP Reviews", where the string "- 3 Reviews" is visually broken 
in two parts), and imagine this string is a link text?
PIZZA <a href="ltr.com">- 3 Reviews</a>
you get a two-parts link:

  3 - AZZIP Reviews
  ---       -------

Unusability added to unreadability.


And what about if "PIZZA" is also a link text?
<a href="rtl.com">PIZZA</a> <a href="ltr.com">- 3 Reviews</a>

  This becomes complicated because the two link texts are rendered as one.
  3 - AZZIP Reviews
  ----=====---------

with rtl.com link (=) between two ltr.com links(-).

Worse, in some browsers (e.g. Mozilla SeaMonkey, others have fixed it 
recently), the rtl.com link is unreachable! (overridden by ltr.com link).

(See here for a simple test 
http://196.200.140.8/Tests/Bidi-Fev-2010/bidiLinkText.html)

Solution:
That's why I suggest to add a § 3.13 to mention the <a> element. I would 
Consider
 link text to be automatically isolated,  i.e. having an implicit bdi 
attribute set to yes.  i.e. making <a ...> <a bdi="yes">.
 
Thoughts?

Najib



Aharon (Vladimir) Lanin wrote:
>  First, a note. The name bdi was meant to stand for "bi-directional
>  isolate". It was not meant to be short for "bidi". I can see how this
>  can cause confusion, so changing the name is probably a good idea.
>
>  Maciej Stachowiak wrotes:
> > I would suggest an attribute name that consists of one or more
> > words. I think a three-letter acronym for a feature that probably
> > won't be used that often in markup will confuse people when they
> > see it.
>
>  A longer name would, of course be more descriptive. However, I do
>  expect this attribute to be used an awful lot by webapps, and do not
>  want to inflate the size of HTML files needlessly. Furthermore,
>  currently existing HTML attributes are generally fairly short.
>
>  Najib Tounsi writes:
> > I suggest going further, in order to permit the user to think in
> > term of meaning of the content [a self-contained entity, e.g. a
> > title or brand name] rather than in term of Bidi "strangeness".
> > Roughly speaking  'sce' should be to 'bdi' what <em> is to <i>.
>
>  Here, I tend to disagree. Bidi strangeness is the only reason for
>  this attribute. To put it another way, what practical meaning would
>  sce have if the bidi problem did not exist? The <em> analogy does not
>  hold: <em>, in principle, could mean that the text is underlined or
>  bolded or have a yellow background. What, besides isolating the
>  content /directionally, /could sce mean?
>
>  Furthermore, to declare a piece of the dom to be something, one would
>  indeed use not an attribute, but an element, as in fact Najib
>  suggests. However, I want to be able to apply the "bidi isolate"
>  functionality to any element - e.g. <br> as in another part of the
>  proposal.
>
>  Thus, I want this to stay an attribute, and an alternative name that
>  come to my mind is dirsep. Any opinions?
>
>  Aharon
>
>  On Mon, Mar 15, 2010 at 3:33 PM, Najib Tounsi <ntounsi@gmail.com
>  <mailto:ntounsi@gmail.com>> wrote:
>
>  Dear all,
>
>  As mentioned in the last i18n teleconf, I suggest to rename the 'bdi'
>  (Bidi Isolation) attribute by 'sce' (Self Containing Entity). Or
>  something like.
>
>  The aim of the new markup (the proposed 'bdi') is to permit the user
>  to define what he/she thinks of as a title, a brand name, and what
>  else, and then should be rendered  visually as it is, i.e.
>  independently from what precede or follow it.
>
>  I suggest going further, in order to permit the user to think in term
>  of meaning of the content rather than in term of Bidi "strangeness".
>  Roughly speaking  'sce' should be to 'bdi' what <em> is to <i>.
>
>  Moreover, and for readability, 'sce' attribute without value seem to
>  me more meaningful than 'bdi' itself. <span sce dir="ltr"> -3
>  review.</span>
>
>  Opinion?
>
>  Regards, Najib
>
>  PS: I even wanted to suggest another element name, to reinforce the
>  idea of bidi isolation. <sce dir="ltr"> -3 review.</sce>
>
>
>  Richard Ishida wrote:
>
>  HTML folks,
>
>  Just to let you know the expectations of the i18n WG wrt this
>  document[1] which was published yesterday.  We do not expect the HTML
>  WG to review and comment on it just yet.
>
>  The document is still in early draft, and was published to facilitate
>  ongoing feedback from bidi experts and i18n folks. It also contains
>  some explicitly identified open issues.
>
>  The plan is to obtain feedback as soon as possible from bidi experts
>  and internationalization folks, then issue a new draft that
>  incorporates the results of those discussions.  Only at that point do
>  we plan to put the proposals to the HTML community and seek their
>  comments and commitment. Depending on the amount of discussion that
>  takes place, we would hope to publish the second draft in about a
>  month from now.
>
>  Best regards, RI
>
>  [1] http://www.w3.org/TR/2010/WD-html-bidi-20100304/
>
>
>  ============ Richard Ishida Internationalization Lead W3C (World Wide
>  Web Consortium)
>
>  http://www.w3.org/International/ http://rishida.net/
>
>  -- Najib TOUNSI (tounsi at w3.org <http://w3.org>) W3C Office in
>  Morocco (http://www.w3c.org.ma/) Ecole Mohammadia d'Ingénieurs, BP.
>  765 Agdal-RABAT Morocco Phone : +212 (0) 537 68 71 50  Fax : +212 (0)
>  537 77 88 53 Mobile: +212 (0) 661 22 00 30
>
>
>
>
Received on Thursday, 18 March 2010 15:53:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 18 March 2010 15:53:39 GMT