[XeTeX] Seeking short examples of complex renderings

[Simon Cozens]

On 06/12/2013 20:18, Khaled Hosny wrote:
> Both Pango and XeTeX use HarfBuzz which in turn can use Graphite, so I
> think HarfBuzz and Graphite are the proper places for these tests (and
> both already have test suites in place)

This is true, but there *are* higher level applications, like XeTeX and SILE 
and Gtk and Firefox and so on...

> Instead of comparing images, which can be affected by things unrelated
> to layout like hinting, it would be better to compare glyph IDs with or
> without glyph positioning, check HarfBuzz and Graphite test suites for
> examples.

...and so I don't think that having low-level tests obviates the need for 
high-level ones.

"My script is meant to look like *this* but your application renders it like 
*that*" is as much a meaningful test as 
f499fbc23865022234775c43503bba2e63978fe1.ttf:U+09B0,U+09CD,U+09A5,U+09CD,U+09AF,U+09C0:[gid1=0+1320|gid13=0+523|gid18=0+545] 
- and possibly more accessible too.

And in fact it's precisely because, say, SILE uses Pango which uses Harfbuzz 
which uses Graphite, it's useful to have an easy way to see who's getting it 
wrong. If SILE messes up a rendering, I want to have some text I can throw at 
pango-view to see if that gets it right. Sorting out the layers is important.

For instance, I see that Harfbuzz already has some tests for Hebrew vowel 
pointings, [*] but running pango-view on these tests produces erroneous 
output. So how should I describe the problem to Pango developers, other than 
by having a picture of what Pango *is* generating and what I think it *should* 
be generating...

...which is basically what I am putting together.

[*] or at least it has some files with some pointed Hebrew in it - but I don't 
see the test suite doing anything with it, nor do I see any expected shapings 
for any of the texts/ directory.