[XeTeX] Discretionary line-breaks in Tamil

[Roland Kuhn]

From a programmer’s perspective it is usually much nicer to not have special cases: if U+200B were just a space (i.e. eligible for line breaks) then the rest would follow normally. Now, it could make sense to retrofit a font during font loading with a trivial definition of U+200B if it does not already provide one.

Disclaimer: I am not familiar with the XeTeX codebase, just commenting from the peanut gallery.

Regards,

Roland

> 30 sep. 2019 kl. 11:57 skrev Suki Venkat <suki.venkat at gmail.com>:
> 
> Exactly!
> That's how all the browsers seems to behave anyway,
> i.e., treating 200B as potential point for a line-break,
> even if it is not defined in the font.
> 
> Suki
> 
> On Mon, Sep 30, 2019 at 4:55 AM Mike Maxwell <maxwell at umiacs.umd.edu <mailto:maxwell at umiacs.umd.edu>> wrote:
> On 9/29/2019 3:02 PM, Suki Venkat wrote:
> > Then went on to hack the hyph-ta.texfile and did "mktexfmt xelatex"
> > to produce nice results using XeLaTeX.
> > It turned out the uni200B was not defined in the font, although uni200C 
> > and uni200D were defined.
> > Then managed define uni200B in fontforge and it does seem to produce the 
> > same result even if the uni200B (ZWSP or DLB) is defined in the font or not.
> 
> I'm speaking from ignorance here--I know nothing of the internal 
> workings of xetex--but it seems to me that the question of defining a 
> glyph for U+200B is beside the point.  It should not, it seems to me, 
> have a glyph.  Instead, xetex should break the line or not when it 
> encounters this code point, and then--regardless of the line 
> break--delete the character.  It's a zero width character, and its 
> height is irrelevant (unlike a strut), so there's no shape to show.
> -- 
>     Mike Maxwell
>     "I am, by a flood, borne back to that wondrous
>     period, ere time itself can be said to have begun;
>     for time began with man." --Herman Melville,
>     Moby Dick

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/xetex/attachments/20190930/0b8cdd52/attachment.html>