- From: Norbert Lindenberg ♻ <norbert.lindenberg@yahoo-inc.com>
- Date: Fri, 16 Apr 2010 20:52:25 -0700
- To: Behdad Esfahbod <behdad@behdad.org>, public-i18n-bidi@w3.org, fantasai <fantasai.lists@inkedblade.net>
- Cc: Norbert Lindenberg ♻ <norbert.lindenberg@yahoo-inc.com>, "Aharon (Vladimir) Lanin" <aharon@google.com>, Simon Montagu <smontagu@smontagu.org>
In a discussion I had with the bidi experts here at Yahoo we came to the conclusion that a) the hflip attribute is not really useful because it works only for very simple unshaded images, b) support in CSS to select an image based on the direction of the context is what's really needed. Fantasai wrote on 2010-03-17 saying that the CSS working group "is considering selectors based on the HTML-resolved base direction of the element, e.g. :rtl and :ltr to style elements differently based on their HTML-declared base directionality." It seems that this feature in CSS would enable our preferred solution. Norbert On Apr 8, 2010, at 06:56 , Behdad Esfahbod wrote: > If this is being pushed in, would need a similar method for flipping > images > defined in CSS. > > behdad > > On 04/08/2010 04:56 AM, Aharon (Vladimir) Lanin wrote: >> Oh, and the value to indicate that the flip should take place should >> then be identical to the attribute name, so it works when used >> without >> an attribute. Thus, I guess it would be: >> >> rtlflip (%URI;|rtlflip|none) # none >> >> On Thu, Apr 8, 2010 at 11:54 AM, Aharon (Vladimir) Lanin >> <aharon@google.com <mailto:aharon@google.com>> wrote: >> >> Should the name of the attribute then be changed to something like >> rtlflip? >> >> >> On Thu, Apr 8, 2010 at 12:48 AM, Simon Montagu >> <smontagu@smontagu.org <mailto:smontagu@smontagu.org>> wrote: >> >> At the recent workshop on the document 'Additional >> Requirements >> for Bidi in HTML' held at the Israeli W3C office, an >> improvement >> was suggested for the hflip attribute, to give authors the >> option of either specifying a separate URL for ltr and rtl >> versions of an image, or that the image should be flipped in >> rtl context. >> >> IIRC, the original suggestion was made by Amit Aharonovitch, >> and >> I strongly support it. Depending on image design, simple >> mirroring for RTL can often produce an ugly effect (for >> example >> with images that use shadows or gradients), and authors may >> well >> prefer to take on the extra burden of maintaining separate >> images for LTR and RTL, but will still want to use this >> attribute rather than having to swap the images depending on >> page direction. >> >> The following is my proposal for the syntax of the attribute: >> >> | hflip (%URI;|flip|none) # none >> | >> | This attribute specifies an alternative image for use when >> the >> | computed direction of the image's parent element is RTL. >> When >> | the value of the attribute is a URI, the URI specifies the >> location >> | of the alternative image resource. When the value is >> "flip", the >> | image should be horizontally mirrored by the user agent. By >> default, >> | the value is "none", which means that the image is rendered >> | identically in both LTR and RTL contexts. >> >> By design, this proposal does not include the possibility of >> an >> original image designed for RTL context, to be flipped in LTR >> context. >> >> >> >> > >
Received on Saturday, 17 April 2010 04:24:25 UTC