Follow-up Comment #2, bug #64728 (project groff):
Yes, the output of groff looks the same as Heirloom Doctools troff/nroff, both
with terminal output and PostScript output. (I think this is what you are
asking for?)
A bit more playing around made me suspect 342–346 in src/preproc/eqn/box.cpp
(in the release version of groff 1.23.0), where it actually looks like the
ending of the lineup equation is aligned with that of the mark equation on
purpose. In case of a lineup equation being longer than the mark equation,
this causes a negative horizontal motion (\h'-72').
Thus, it seems the bug lies indeed deeper. The following example shows a
diversion in which 'abc' is printed, followed by a negative horizontal motion
of the length of the just printed. The reported diversion width is zero; the
expected width would be \w'abc'u, as that's the width of the conceptual box of
affected output. Had this not been a diversion, but a string, then a width of
zero might be desirable (though this doesn't seem to be so well-defined in the
manual).
$ cat width_negative_horizontal_motion
.di xx
abc\h'-\w'abc'u'
.br
.di
Diversion: \*(xx
Width: \n(dl
$ nroff width_negative_horizontal_motion | sed '/^$/d'
Diversion: abc
Width: 0
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?64728>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
