Re: [Groff] bug in GNU troff?

[Ralph Corderoy]

Hi Werner,

> > That doesn't matter.  groff was meant to be "a complete, open
> > source implementation of the standard Unix typesetting system" --
> > James Clark.
> 
> In compatibility mode, yes.  But, sorry if I say so, this particular
> behaviour of UNIX troff is (at least for me) neither intuitive nor
> very useful

This list's members don't seem to fullly represent the troff/groff
usership;  consequently, the absence of a list reader saying `I use it'
doesn't mean there aren't people that do, especially since it is
documented and well known.

> I'm really interested to see a practical application.

First stab, how about this.

    .ta 1iR +1i
    Margin  Right   Left
    .br
    x       x       Positive
    .br
    x       x\c
            Negative

It gives

    Margin     Right          Left
    x              x          Positive
    x              x  Negative

Is that what you mean?

> > There's a danger of trying to coerce groff to what you think troff
> > should have been.
> 
> Which danger?  Please elaborate.  Since the very moment of
> introducing long names, groff is no longer compatible with UNIX
> troff.

True, but there seems to me to be a difference between *extending*
groff with long names, although incompatible with troff's parsing in a
few cases, and re-defining the meaning on a tab in the input.  I agree,
it is a question of judgement, but changing the meaning of tab on input
just seems a step too far.

Given that, when explicitly questioned on this, Ossanna insisted on the
behaviour to his colleagues, who were heavy troff users and top brass
to boot, I think we should accept his argument of why it was the Right
Thing for troff.

Compatibility mode is too coarse a granularity.  If on then long names
are sorely lacking.  A set of compatibility flags to control each group
of extensions might help.

> I don't want to be a guard of a museum.

I sympathise;  the nights are quiet here at the A&V.  But there is
something to be said of keeping *troff* alive and well and useful in a
modern environment.  Does TeX suffer from being a museum?  Am I right
in thinking Knuth has fixed its definition and therefore freedom to
tweak and twiddle is limited?  Troff's definition is pretty fixed too,
and wandering away from it might be unwise.


Ralph.