Quoth G. Branden Robinson:
My questions/challenges for the reader are:
A. In which of the five do you expect to see "ABC" emboldened?
B. In which of the five do you expect to see the special characters
emboldened?
Groff 1.22.4, plan9port troff, Heirloom troff, and 2.11BSD¹ troff
all behave the same: 2, 3, and 4 all have ABC and the characters
from the special font emboldened; in 1 and 5, nothing is bold.
The question is how emboldening a normal font affects special
characters. CSTR #54 says (§ 2.2p3):
characters on that [special] font are automatically handled as if
they were physically part of the current font.
With that context, I think the behavior observed makes sense. The
characters in special fonts are handled as if part of R, so they,
too, are emboldened when R is.
I wouldn’t be too surprised about a different interpretation: One
of not emboldening the characters taken from the special fonts if
just R is emboldened by .bd. Just emboldening the characters from
the special fonts if S is emboldened, or even just if the second
form of .bd is used. I couldn’t tell, and guessed wrong (that the
characters from the special fonts wouldn’t be emboldened at all).
I lose.
CSTR #54 makes for a great reference. It is not perfect, and I’m
all in favor of having better and better references.
I am happy that the results are consistent in all the troffs
I tried.
¹ On early Unices, troff -t gives you the C/A/T input and tc gives
you Tektronix 4014 input from it; XTerm can simulate a
Tektronix 4014.
