troff:ono.tmac:289: warning: register '.meta' not defined [-w reg]

[g-branden-robinson]

$ nl ono.tmac | sed -n 267,302p
   267  .de ``
   268  '       \" Begin block
   269  '       ie !\\n(`` \{\
   270  .               br
   271  '               ie \\$2>1 'nr `` \\$2
   272  '               el        'nr `` 6
   273  '               in +\\n(``n
   274  '               de ` >>
   275  '                       in -\\n(``
   276  '                       ft \\n(.f
   277  '                       if \\n(.u 'fi
   278  '                       hy
   279  '                       lg 1
   280  '                       do hy \\n[.hy]
   281  '                       do lg \\n[.lg]
   282  '                       do kern \\n[.kern]
   283  .               >>
   284  '               nf
   285  '               nh
   286  '               lg 0
   287  '               ft CW
   288  '               do kern 0
   289  '               do if \n[.meta]=1 \X'meta: begin code \\$1'
   290  '               tr `\`'\*'^\*^~\*~-\*-
   291  '       \}
   292  '
   293  '       \" End block
   294  '       el \{\
   295  '               tr ``''^^~~--
   296  '               do if \n[.meta]=1 \X'meta: end code'
   297  '               `
   298  '               rm `
   299  '               rr ``
   300  .               br
   301  '       \}
   302  ..

It looks like you forgot to escape the interpolations of the .meta register within the macro definitions; they are being evaluated at package load time.

My suggestion:

. do if \\n[.meta]=1 \X'meta: begin code \\$1'
. do if \\n[.meta]=1 \X'meta: end code'

Also, I would not spray and pray with the no-break control character as seen above. The GNU troff manual cautions against the practice, and line 270 suggests that prayer can fail. I suggest using it only with the handful of requests the no-break control character can effect. A future version of groff might issue a warning in category style about it; see comments 4 and 28 to Savannah #62776. You might of course wait until the style warning is implemented, and then use it to find the occurrences for you, obviating a tedious manual review.

[Alhadis]

It looks like you forgot to escape the interpolations of the .meta register within the macro definitions; they are being evaluated at package load time.

Actually, I didn't. See, the .meta register is part of a yet-to-be-reifed idea for a Troff preprocessor to demarcate regions of semantic importance by injecting \X'meta: … before and after landmark tags in well-known macro packages. The .meta register is intended to be defined when the package is loaded, serving as the moral equivalent of Groff's -P arg switch to pass options to a postprocessor.

Also, I would not spray and pray with the no-break control character as seen above. […] A future version of groff might issue a warning in category style about it

Alright, now this is a more serious issue. The no-break control character is expected to be used to mask the .[ hyperlink macro when preprocessing with refer(1) (which will complain about malformed reference blocks, seemingly ignoring the unfamiliar parameters following .[ that should clearly tip refer(1) off that this isn't a bibliographic delimiter, but something else entirely.

But I guess I'll fix the frivolous use of non-break control characters in Mono itself for the time being. I'll burn the biblio-bridge when I get to it…

[g-branden-robinson]

Also, I would not spray and pray with the no-break control character as seen above. […] A future version of groff might issue a warning in category style about it

Alright, now this is a more serious issue. The no-break control character is expected to be used to mask the .[ hyperlink macro when preprocessing with refer(1) (which will complain about malformed reference blocks, seemingly ignoring the unfamiliar parameters following .[ that should clearly tip refer(1) off that this isn't a bibliographic delimiter, but something else entirely.

But I guess I'll fix the frivolous use of non-break control characters in Mono itself for the time being. I'll burn the biblio-bridge when I get to it…

There are no plans to issue that warning with respect to macro calls. Only request invocations. Warning in the former case would discourage or defeat exercise of a GNU troff selling point, the .br register.

If you have a macro named [, and call it with ', refer(1) will ignore it.

[Alhadis]

There are no plans to issue that warning with respect to macro calls. Only request invocations

Ooooh, right. Yes, that would make sense. 😅 In that case, ignore the second half of my response.