Re: [Groff] OT: \} considered as macro argument regarding register .$

[Carsten Kunze]

Hi Ralph, 

> I can't spot anything to avoid \n($. counting \} from browsing the 7th
> Ed. source.

Maybe this is because the parser should separate the terminal symbols belonging 
to a conditional block from statements inside the block. So when the macro call 
is parsed the \} should not be there anymore (and .$ does not need to account 
for it).

>  Note, 7th Ed. has been ported to x86 and can be run as a VM
> if you want to test it.  http://www.nordier.com/v7x86/

I did not get it to run in short-term, but I'll keep at it, thank you!

> BTW, as well as \n(.$ including \} in its count, it naturally follows
> that \$3, for example, is \}.

It would be consistent but not desirable. And fortunately it is not the case. 
More often then .$ I use ".if ^G\\$3^G^G ..." and this works in this case (\$3 
is empty).

> Agreed.  But documented to avoid surprises.  I checked CSTR 54 and
> couldn't find anything covering this.

Yes, Werner had suggested this also and I reported the issue (that this 
behavior should be documented).

I had expected that it is not mentioned in CSTR54. If this issue had been there 
already, J. Ossanna or B. Kernighan had not been aware of it--else they had 
fixed it (I am quite sure that it's a bug and not intentional.)

Cheers,
Carsten