Glyphs sometimes don't render with unicode hexcodes and pdf/ps/html devices

Hi,

If I make a document that contains the single line:

   \f[U-S]\[uF8EB]\f[]
   
to get the "" glyph and then convert the document with

   groff -Tpdf -k -K utf-8 -dpaper=A4,

only a blank page is produced. Same thing happens with the "ps" and "html"
devices.

The "utf8" device shows the character fine.

If instead I use the mnemonic alias for it like: 

   \f[U-S]\[parenlefttp]\f[]
   
then it renders okay with both pdf, ps and html.

This only happens with some glyphs, but the solution is not to use
the aliases, generally, because that fails in some other instance.

Do you see the same and what is your take on it?

Regards,
Morten