> Input:
>                       \-              -
>                    MINUS SIGN       HYPHEN
>
> Output with the current groff CVS:
>
> groff -Tutf8        U+2212          U+002D
>                    MINUS SIGN     HYPHEN-MINUS

This is fixed now; it will again create HYPHEN instead of HYPHEN-MINUS
(and this can be overridden on demand as documented).  The solution
was quite simple: For the UTF-8 device, I now say

  .char - \[hy]

which makes `-' a pure input character, quite similar to the soft
hyphen.  As far as I can see, this is the only character in the range
0x00-0x7F which has to be handled specially, right?  Everything else
map to itself even in UTF-8 mode.

> groff -Thtml        U+2212          U+002D
>                    MINUS SIGN     HYPHEN-MINUS

This stays as-is.

> groff -Tdvi,        U+002D          U+002D
> dvipdf, acroread5,  HYPHEN-MINUS   HYPHEN-MINUS
> copy & paste

Using dvipdfmx instead of dvipdf (from TeXLive), I correctly get MINUS
SIGN and HYPHEN-MINUS.  AFAIK only dvipdfmx supports ToUnicode maps.

> groff -Tps,         U+00AD          U+002D
> ps2pdf, acroread5,  SOFT HYPHEN     HYPHEN-MINUS
> copy & paste

Well, using ps2pdf from ghostscript 8.55, xpdf correctly gives MINUS
SIGN and HYPHEN-MINUS.  Additionally, acroread 7.0.8 (on my GNU/Linux
box) returns the same.  One of the tools in your change is apparently
broken...


    Werner