See also http://tex.stackexchange.com/questions/86565/drawbacks-of-xspace with the first answer from the package author of \xspace. Quote "So, if you find it useful, fine, it's there. But personally I wouldn't recommend it."

I've added a separate commit to this pull request that removes \xspace from the definition of \grammarterm. A diffpdf in "appearance" mode shows 239 differences. I've looked through the first ones and found lots of noise and mildly reduced whitespace between a \grammarterm and a following punctuator as the only significant difference. For a full stop (period), this actually seems to improve matters; there was awkward extra spacing (just a tiny bit) between the r and the period in
\grammarterm{identifier}.

I'm not prepared to take a macro change the day before a working paper goes out unless someone has looked at all the resulting formatting changes; a survey of the first few is not sufficient at this point. I'll take this if either you've looked at all 239 and they are all OK, or if you change the "\grammarterm{cv-qualifier}s" to "\grammarterm{cv-qualifier}{s}", or even if you don't add the \grammarterm at all (since it was a drive-by anyway). I'm also a bit surprised we see any differences here in cases where \xspace was not adding a space; do we really have 239 bogus spaces inserted by this, or does \xspace also somehow change the formatting even when it doesn't add a space?

If we don't make the macro change now, I'd still like to make it at some point. Generally I'd like for us to get rid of the \xspaces entirely. (See http://tex.stackexchange.com/questions/86565/drawbacks-of-xspace for the reason why, and note that even the original author of the package does not recommend it.)

@zygoloid I wouldn't be surprised if \xspace interferes with the kerning between two characters (i.e., preventing kerning altogether). This would result in minute spacing differences.

@godbyk Thanks, that makes sense.

I don't think that changes my position, though; I'd like to check that none of the 239 changes are losing a space that we want; that could happen pretty subtly:

the \grammarterm{foo-name}%
\indextext{...}%
is the name of a foo

... would I think lose the space after foo-name with this change. I can check this myself if no-one else does so first, but I have a few other things I need to get done before the pre-meeting mailing deadline.

@zygoloid To clarify, my comment wasn't intended to change your position—I agree with it, given the inconsistencies I've seen in the code—but rather to provide an explanation for some of the minor changes you may see.

Updates:

• Removed change to macros.tex; this is handled by issue Review use of \xspace in macro definitions #1433.
• Used \grammarterm{cv-qualifier}{s}