Shavian text and other 4-byte utf8 fonts corrupts on display after several characters.

Shavian text is displayed unreliably after several non-identical characters.

The attached files demonstrate corruption as witnessed between Arora browser and Firefox of a page using 4-byte UTF8 Shavian (Shaw) English characters.

Text typed directly into a QLineEdit (or any text input location) is also affected.

If the same shavian character is entered repeatedly into an entry, the corruption of the text as displayed does not appear to occur.

Once the text is corrupted by using various Shavian characters, I could produce a stream of latin "u" characters by holding down my shavian-a key, then the latin "l" (lowercase L) by holding down my shavian-r key. Typing random characters after this restored a sane stream of characters eventually, but only briefly.

(If needed, Shavian fonts can be found at http://marnanel.org/shavian/fonts/ - I used Androcles and Andagii in my testing.)

Upd. Similar problem appears not only with Shavian fonts, but with any font from the 4-byte range. I have a problem with mathematical (bold, itallic, etc.) fonts. D. Vlasov.