Re: [CSS3-UI] text-overflow change: single value must only apply to end line edge

On 29/09/2011 6:54 PM, Tantek Çelik wrote:
> Through experience with both existing site use of text-overflow, and
> testing existing implementations[1], it's become clear that
> text-overflow with a single value must only apply to the end line
> edge, rather than to both left and right line edges.
> In particular, there are numerous sites that apparently depend on
> start edge ellipsing NOT occurring when there is a single value (e.g.
> cases where a negative text-indent is combined with left padding, thus
> causing overflow of the text, but not any rendering clipping).
> In addition, existing implementations (IE, Opera, Safari) consistently
> apply a singular text-overflow value to only the right edge for LTR
> text, and in most of those (IE, Safari) to only the left edge for RTL
> text.

I would say this is far worst than you recognize (with regards to 
overflow, WebKit and Gecko are more in agreement). BTW, LTR and RTL is 
'inline base direction' and is not exclusive to text. It encompasses 
'inline base direction' and has a direct relationship with 'block flow 
direction' (was 'block progression'). Please see the introduction of the 
writing-mode WD [2] for more details.

> Thus I've made the respective change in the CSS3-UI text-overflow section:
> from: a single value applying to both left and right edges
> to: a single value applying only to the end line edge.
> Thanks,
> Tantek
> [1]

Looking at the comments in that bug thread (epically by Boris [3]) and 
with my own experimentation with ellipses, bidi, overflow (both 
direction and block progression, the former is text overflow), I would 
say that it is not an issue with ellipses but more an issue with 
overflow of any nature. To fix ellipses handling on a broken overflow 
behavior will just cause more compat issues in the future.

Please view this test case.

For the above test case, with the earlier examples, there seems to be 
either WebKit/Gecko behavior or Opera/IE behavior. With the later 
examples, Opera seems to have WebKit/Gecko behavior leaving IE alone 
with it's behavior.

May I suggest that all CSS modules, implementations and the CSS WG start 
to work towards a CSS3-overflow module instead of a collection of notes 
[4] for planning, so overflow can be defined in one module (not defined 
in a myriad of ways in many modules) so we can work towards 
interoperability between all UAs.

Please see more in a future list message regarding a new property that I 
like to propose which is 'overflow-direction'. This is needed if UIs 
consider showing bidirectionally of script or dual writing-modes 
(horizontal and vertical) in a sensible way.


Alan Gresley

Received on Thursday, 29 September 2011 12:53:29 UTC