Re: Fribidi and joining

[Roozbeh Pournader <roozbeh at sharif dot edu>]

On Thu, 22 Aug 2002, Mohammed Elzubeir wrote:

> This is already the case (non-uniform displayed text). 

Well, this is the current case, but will hopefully not be the case very
soon (say, two years from now).

> > Put these in a can with line breaking (which need to be aware of the width 
> > of the letters to find the appropriate line breaks), and you'll be in a 
> > complete confusion.
> 
> Okay, so why aren't we simply ignoring control characters, bidi formatting
> codes, etc when doing the shaping (post-bidi)?

We can't ignore control characters like ZWJ and ZWNJ when we are doing 
shaping. They are *shaping* controls.

> Looking through the ICU pages I can see they have _something_ about shaping,
> but it's not clear to me whether that is tied to ubidi at all.

But they don't have a clear solution with our case.

> The ICU library seems to have a reverse RTL function: [...]

Sure, I think I can also write one. But the output of such a function will
be just one of the many logical character sequences that render like the
input. In other words, both me and you can have such functions, and be
compliant to Unicode, but the functions may have different outputs when 
given the same input.

Gaspar Sinai was asking about a reversible bidi *algorithm*.

roozbeh