Hi Tadziu, > > > > \l'\n(.l/1m' > > > \l'\n[.l]u' > > I think what should be considered here is that groff only does integer > arithmetic. Therefore, in the first case the length of the line will > get truncated to an integer number of ems, whereas in the second case > the length of the line will match the width of filled text lines as > closely as possible within the limits of the device resolution. Thus > I suggest that the second way is generally preferable.
Agreed. And my default of reaching for nroff to test isn't great because its 240 units per inch breaks down into 24 per em *and* 24 per en. Also, setting the line length with .ll applies rounding to devascii's ‘hor’ that is 24, the same as an em's units, so rounding issues with the division don't show up. -- Cheers, Ralph.
