Comment # 18 on bug 1006510 from Petr Gajdos

(In reply to Fuminobu Takeyama from comment #17)
> Since now Ruby keeps encoding with string object. As the default encoding to
> read a file is ASCII-8BIT, the code reading family name from fontconfig
> maybe push the family name into a Ruby string object in ASCII-8BIT. At that
> time, the byte array of the string is the same with the appropriate string
> object. So force_encoding works fine.
> 
> Is it the only place that needs force_encoding?

Hard to say when I do not know the exact cause; other parts seem not to have a
problem with family strings (we take family strings only from the fontconfig).
So I guess the problem is there only when strings are connected via gettext
(big question mark here, but see comment 0).

And question is what happens when .force_encoding("utf-8") is applied on non
utf-8 string (I had not verify we have such font).