Re: [Groff] Re: underlining in nroff.

[Werner LEMBERG]

>     <1>.<1> <Section> <Heading>
> 
> where <> marks the underlined text.  So \fI is .ul.
> 
> But, you see how the `.' in 1.1 isn't bold with .ul?  _Introducing
> Unix System V_ says .ul only underlines alphanumeric characters and
> you need .cu to give continuous underline on all characters,
> including space.  That seems to match the `.' not being bold.

But cstr54.ps (which is the definitive guide, I think) only talks
about an `output-device-dependent subset of reasonable characters'.
It is highly debatable to define `reasonable'.  So I don't see a
reason to change groff's current behaviour.

If there is really some desire to imitate Unix troff's behaviour I
could introduce another property to be set with the `.cflags' request
(it is locale-dependent to decide what `alphanumeric' really is).  I
don't think that it is too difficult to implement, but this isn't
really something urgent, isn't it?

> With `groff -Tascii' I get the same output for all three

Use the current snapshot, and you'll get something different with .cu.

> Furthermore, it explicitly says troff (meaning ditroff) is the same
> as nroff in its `alphanumeric only' treatment but that the
> difference is normally only noticable when characters such as `['
> are involved because the italic version is clearly different.
> `groff -Tps' again gives italics for all characters with .ul.

Yep.

> This is really just to throw up more historical material and testing
> for discussion rather than a firm view as to what should be done.

Thanks for the info.  I'll put it eventually into the compatibility
section of groff.texinfo.


    Werner