Re: [Groff] groff_ms.man

[Werner LEMBERG]

> 1. Each device can provide its own implementation of ellipsis, or
>    anything else, and this can be referred to in groff in a
>    device-dependent way.  The macro files tmac.ps, tmac.dvi, tmac.X,
>    tmac.lj4, tmac.latin1, tmac.tty, etc. already do this more or
>    less extensively.  I don't think "device compatibility" is an
>    argument here, though I can see the point of a default for
>    devices which don't have a tmac.<dev> file, nor a built-in
>    ellipsis.

My guide for providing a default definition for the ellipsis is
LaTeX.  This is glyph is *not* device dependent!

> 2. For some time recently I have formed an increasing perception
>    that the formatting of man pages is becoming an unduly dominant
>    influence on groff.

Mhmm, not really.  I always try to test everything with non-TTY
devices also.  On the other side it is correct that until now I've
revised the macro packages for man pages only (mainly because there
were simple to understand for a beginner like me), fixing a lot of
things to improve the appearance on TTY devices.  But this won't
decrease the quality the layout on non-TTY devices.  If you have this
feeling please describe the exact effects.

>    What is specifically needed for formatting man pages should be
>    wrapped up in tmac.an and associated files.

This has been always true and will remain so.

>    I do not agree with Alejandro's implied argument above
>    ("therefore it can be used for man pages") that this purpose
>    justifies setting a particular definition as a default, when for
>    some devices there is a better, built-in one.

Hmm, until now I haven't seen a built-in ellipsis glyph which is
suitable for all purposes on a particular device.  As mentioned
earlier, the ellipsis glyph from the Type1 Symbol font is limited in
its use.

> 3. If by "the ellipsis proposed by Werner Lemberg should be the one
>    supplied" you mean the suggestion
> 
>       .char \[...] .\|.\|.
> 
>    then this is OK for typewriter-like devices, since \| has no
>    effect (but then the definition is unnecessary since it is
>    equivalent to "...").  It is not needed for PS (and indeed
>    produces a different effect from \(el),

???  groff has no `el' glyph!  Nor does any macro package define such
a character or macro.  It seems that you are talking about a private
extension in your macro/glyph set.

> Where there is an available ellipsis in the device, then let it be
> used unless it is unsuitable for a particular purpose (in which case
> define something that is).  If it is not available, then let it be
> defined in the device specific tmac.<dev> file.

Oh, this reminds me that the correct definition will be

  .fchar \[...] .\|.\|.

Sorry for the confusion.  If a particular font defines a glyph for
\[...], it won't be overridden by the character defined with fchar.

> The proposal to put such a definition in the troffrc file makes it
> the default for all devices unless over-ridden in the tmac.dev file;
> as I argue above, this is the wrong way round in my opinion.

I don't think so, and I hope that after clarifying the use of `fchar'
you can agree with it.

> Nor do I agree that "It's a rather useless glyph IMHO"
> [Werner]; I use it extensively ... !

Try the following with the PS device:

  .char \[ell] \f[S]\N'188'
  .char \[...] .\|.\|.
  .ps 50
  .vs 20
  .ft B
  This \[ell] and that \[ell]

  This \[...]\& and that \[...]

  This ...\& and that ...

This is why I say that the ellipsis glyph in the Symbol font is
useless as a \textellipsis (to use the corresponding LaTeX macro).
Its use is restricted to mathematical typesetting.  It also shows why
there must be some space between the dots.


    Werner