Hi,
All of the direction-unicode-bidi testcases rely upon a reference image,
with a pass condition of "test passes if characters are in the same
order", and use a mixture of English and Arabic (presumably to have one
ltr and one rtl language).
However, in some cases, it's not immediately obvious whether some
testcases pass; for example, in the screenshot attached it's not clear
to me whether the first/last Arabic characters are the same as in the
screenshot, but it could just be a difference of fonts/ligatures.
It'd be desirable to have the testcases not being reliant upon a
screenshot (which inherently will only be exactly equal for a specific
rendering of a single font), but at the very least minimizing font
differences (not having the tests being affected by ligatures would
help, I think).
--
Geoffrey Sneddon — Opera Software
<http://gsnedders.com/>
<http://www.opera.com/>
