Re: [Groff] meaning of \[or] glyph

[Werner LEMBERG]

> > The problem is that \[or] might have been used for different
> > purposes, e.g. denoting divisibility of whole numbers.
> 
> But not bracket building?  I know it isn't used for that these days,
> but perhaps historically.

According to the original troff documentation, the proper glyph for
this purpose is \[bv].

> > Anyway, I have to make a decision, and I have decided to be
> > conservative (contrary to my own conviction): groff will treat \[or]
> > as font-invariable, i.e., the font shape will always be the same.
> 
> If it's of any interest, the output of
> 
>     .nf
>     .de x
>     .ft \\$1
>     \\$1 abc\\(ordef
>     ..
>     .x B
>     .x BI
>     .x C
>     .x CB
>     .x CW
>     .x H
>     .x HB
>     .x I
>     .x R
>     .x S
>     .x TB
>     .x TD
>     .x TI
>     .x TR
> 
> on AIX 3.2.5's old `troff -Tpsc' shows \(or as the `|' character,
> e.g.  ASCII 124, from the Symbol font in all cases except C and CW
> where it is the C and CW's `|' character.

Sigh.  I won't emulate this.

To make life simple (well, the whole mess is already too difficult to
be called simple -- it costs way too much time to fix it), I will only
support two categories of glyphs: font-variant and font-invariant
ones.  The former change the shape according to the current font
(provided a glyph does exist), the latter will always be taken from
font `S', or a definition with `.char' is provided.  And I will
document which glyph belongs to which category.


    Werner