Re: [css-text] Arabic letters connecting between elements with display: inline

On 5/26/2014 12:51 PM, Behdad Esfahbod wrote:
> On 14-05-22 06:06 PM, Asmus Freytag wrote:
>> On 5/22/2014 2:43 PM, Behdad Esfahbod wrote:
>>> On 14-05-22 05:24 PM, Asmus Freytag wrote:
>>>> What software support joining (other than color) today?
>>> Android platform (not the browser), and GNOME desktop both do.
>>>
>> Can you share a screen shot?
> Here are the slides of my 2004 Unicode Conference talk.  See page 16 for example:
>
>    * In the slides, I've used color to highlight letters, while preserving shaping:
>
>    http://behdad.org/download/Publications/persiancomputing/slides_acro.pdf
>
>    * In the print version, I've used bold font to do the same:
>
>    http://behdad.org/download/Publications/persiancomputing/a007.pdf
>
> Surely not the most typographically correct way to do it.  But anything other
> than preserving joining would have been just wrong.

A couple of comments:

Looking at your example, the version with bold fails miserably. The text 
of the presentation is in English, making clear your intent of 
communicating with people that are not native users of the script. The 
color distinction is readily grasped, but the weight distinction is not 
apparent, to the untrained eye the weight contrast disappears entirely. 
Unlike typical examples in Latin, the type color is not noticeably 
darker. In fact, I had to look at the colored example to locate the 
breaks so I could verify that you had indeed used different font weights.

The same applies to the Pango example. If not for the noticeable break 
in stroke thickness, and the high magnification, the effect is easy to miss.

My conclusion would be that this feature is of limited practical 
utility, assuming the outcome of your efforts is typical.

This is more of a question: what tool(s) did you use to create these?

Forcing everybody to implement something that is of limited utility 
doesn't seem the right answer, somehow. I can see why an expectation 
that color (or underlining) should not affect shaping is both reasonable 
and useful. And for those examples, it seems that not only would this be 
useful to authors and readers of such sites, but the implementations 
should have no difficulties arriving at outcomes that are "predictable" 
by authors.

For generalized font and style breaks, I think it would be enough to 
recommend that these are not intended to break shaping runs. I don't 
know whether the results can be made comparable across implementations 
in the general case (radical differences in font family or size may be 
handled differently). It would not seem beneficial to pretend that this 
would simply "work" for the general case.

How would you handle the lam-alif ligature, if it had a font or style 
break between the two letters?

A./

Received on Monday, 26 May 2014 20:21:15 UTC