nroff mode paging inconsistencies

[Chuck Jungmann]

A macro I wrote exhibited rare misaligned formatting. While debugging 
it, I discovered some inconsistencies with page dimensions and offsets 
that I didn't expect from nroff output to a pager.

The alleged (maybe I don't understand something) inconsistencies pertain 
to values of various registers.  The register for page offset (\n[.o]) 
always reports a 0 value, as one would expect for an unpaged document.  
I eventually figured out that I was near the bottom of a page by using 
the .mk action, which does report a vertical offset from a given page 
(though not the same value as \n[.o] in a document formatted with 
troff).  Comparing the .mk value with the page height register (\n[.p]), 
I could see that my formatting problem occurred when I attempted to .sp 
up to a line that was printed before a phantom page break.

Using .mk to isolate the problem also suggests a work-around. Use the 
value reported by .mk in place of the page offset register.  Easier 
still would be to detect nroff mode and if in nroff, just issue a .bp 
before every multi-line object since .bp seems to have no effect on the 
output beyond resetting the position reported by the .mk action.

I sent this bug report to make a record of the issue in case it helps 
someone diagnose and work-around a similar problem.  I'm not sure if 
this is something you would work on.  In reading the 10-year-old mission 
statement, this issue doesn't apply to either of the two areas of 
development.  If you want more information, I would be happy to provide 
more information and a simplified roff-script to manifest the problem.

I'm not sure how you'd fix it, and maybe you can't.  Having a page 
offset from the top of the document risks integer overflow, especially 
with troff output where the my pdf line height is 12,000 basic units, so 
I can see an argument for having virtual pages to limit the internal 
page offset value as reported by .mk.