[sword-devel] Unicode normalisation and arabic vowel filter

[Peter von Kaehne]

Guys, I wonder if someone could give me a piece of advice on a problem
which bugs me. Enormously.

We do have an Arabic vowel filter which is essentially the same as the
Hebrew points filter. In fact they are so similar that Chris at one
stage wondered whether to amalgamate them and throw into it all other
diacritics one might want to filter too.

But - it actually does not work on any Arabic modules published and
privately available.

I.e. the diacritics do not vanish.

I am not sure what is going on. My understanding is that a text needs to
be NFC normalised to be accessible to diacritical stripping and that is
done AFAICT.

How can I debug this?

Peter