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

On 5/27/2014 9:55 AM, Richard Ishida wrote:
> On 27/05/2014 17:43, Asmus Freytag wrote:
>> I think inserting markup between parts of an obligatory (or intended)
>> ligature is a mistake by the author and requiring complex support to
>> force this to "work" is counter productive.
>>
>> On the contrary, the expectation should be clear that ligatures are
>> broken, so that authors can get visual feedback and proof their work.
>
> The problem I see with this is that one of the many possible ligatures 
> like lam-meem depends on the font used.  If the content author changes 
> the font or a device changes the font from one that doesn't support 
> lam-meem ligatures to one that does you don't want to have to go 
> change all the markup or disable the effect.

What I am saying is that if you break the text by inserting markup that 
you are guaranteed that you will not get a ligature there. That way, 
things become predictable again. For example you can color each part. 
However, the results may not look typographically the best. Just as you 
would not color [Of][fice] if you intend to allow the font to use an 
"ffi" ligature.
>
> This may however been a different kettle of fish from a lam-alif 
> ligature, since the latter makes it more difficult to separate which 
> part of the glyph represents which underlying character (at least in 
> the form that looks like a crossed ribbon). I suspect, though I 
> haven't tested it, that in general non-lam-alif ligatures such the 
> lam-meem type keep the glyph representations of the underlying 
> characters separate enough to allow differential colouring.
Since there are no metrics for partial glyphs, this can be approximate 
at best.
>
> A quick test on Firefox shows that it breaks a lam-alif ligature if 
> there is markup between the lam and alif, but that it tries to 
> highlight the right parts of a lam-meem ligature.  IE, however, didn't 
> do so.

Highlighting partial ligatures during editing is a good edge case. For 
highlighting the goal is to give visual feedback that there are multiple 
underlying characters in the text; it is less important to get the 
"correct" part of the ligature, in fact, highlighting simply a half (or 
third, or whatever fraction) is quite workable.

But if you intended to pick out the contribution of each letter in the 
ligature by color - that can't be done in the general case, and having 
the spec set up that expectation would be counterproductive.

In the general case, ligatures are free to be completely fused forms; 
before it became a character, something like "&" was a ligature (of 
"et") and there's no way to apply style to any part of it.

Weight, font-family, size etc. can't be applied to parts of a ligature 
even a "well behaved" one like the ffi ligature.

A./

Received on Tuesday, 27 May 2014 17:26:27 UTC