[HarfBuzz] harfbuzz-ng: Branch 'master'

[Tom Hacohen]

On Thu, 2011-05-05 at 13:11 -0700, Behdad Esfahbod wrote:
>  src/hb-buffer.cc |   16 +++-------------
>  1 file changed, 3 insertions(+), 13 deletions(-)
> 
> New commits:
> commit 3935af1c0d0f53a5fd6054e1ee219f3adda42dca
> Author: Behdad Esfahbod <behdad at behdad.org>
> Date:   Thu May 5 16:09:45 2011 -0400
> 
>     [buffer] Remove wrong optimization
>     
>     While the cluster fields of the glyph string are usually sorted,
> they
>     wouldn't be in special cases (for example for non-native
> direction).
>     Blindly using bsearch is plain wrong.  If we want to reintroduce
> this
>     optimization we have to make sure we know the buffer clusters are
>     monotonic and in which direction.  Not sure it's worth it though. 

If you ever mark monotonic (I guess direction is already known because
of bidi direction - or are there any other cases for monotonic?) please
also expose it. For the same reason I linear search and not binary
search the clusters.

If "in-run" reordering is only possible in a given number of scripts, we
can quite easily mark if it's monotonic or not (or more correctly give a
good enough approximation). We should mark everything as monotonic,
expect for the few scripts who can be non-monotonic, which are, from my
understanding, only the Indic scripts, but this could be for any
reordering shaping engine. Is this correct?

--
Tom.