Pango modules Hebrew heurisics

[Dov Grobgeld <dov grobgeld gmail com>]

While playing around with pango rendering did I notice a regression in the placement of Hebrew diacritics marks for fonts that lack a GPOS table. In my old Hebrew pango module, I had added some heuristics for how to do such placement, and the result was quite acceptable for most fonts. From what I understand from the logs the language specific modules were discarded when Harfbuzz was introduced as a pango dependency (in 2012), and with it my  placement heuristics.

The question is if this was intentional, and whether you consider this is a "font problem" and not something that the pango should deal with?

Regards,

Dov