I've been frustrated by groff's absolute spacing (.sp |<dist>) for decades.
First off, <dist> is calculated from the first baseline of the vertical spacing currently in effect when .sp |<dist> is called, not from the top edge of the printer medium or PDF/PostScript display. The behaviour is documented in the info manual and a workaround is provided, but that doesn't obviate that while for tty devices the behaviour makes sense, it doesn't for -Tps and -Tpdf, which, I posit, are more frequently used in 2021 than tty devices, manpages notwithstanding. Secondly, this behaviour has consequences beyond the annoyance of having to add -\n(.vu or -1 to absolute spacing requests. Consider: .ds A "Now is the time for all good men .ds B "to come to the aid of the party. . .\" First half .nf .ps 12 .vs 24 .sp |1i-1 \*A .ev foo .evc 0 .vs 18 .di bar \*B .di .ev .sp |1i-1 .in \w'\*A 'u .bar .in . .\" Second half .sp |2i-1 .mk \*A .ev foo .di baz \*B .di .ev .rt .in \w'\*A 'u .baz The first half uses '.sp |dist-1' before outputting its diversion, the second uses .mk/.rt. In both cases, processed with -Tpdf or -Tps, string B is higher than string A; diversions whose vertical spacing differs from the prevailing one are not output at the location arrived at by '.sp |dist-\n[.v]u' but rather '.sp |dist-\n[.v-at-top-of-diversion]u'. In a demonstration like this one, the problem is correctable by transparently embedding the .ev foo/.evc 0 requests in the first half inside its diversion, but in complex, trap-invoked macros this strategy may not be desirable or even possible. Alternative, context-sensitive solutions are often non-evident and fussy to implement. I believe groff needs a dedicated request for unambiguous absolute spacing. One that says, "Advance from the top of the printer sheet to the location I say and print whatever comes next right there." The same goes for .mk/.rt. (Preparing mom 2.5, I was surprised to discover that .rt suffers from the same problem, one of the reasons I've decided to bring this up.) What say ye? I'd like to put this in as feature request, but there's not much point if others don't see its utility. -- Peter Schaffter https://www.schaffter.ca
