Re: [Groff] Re: Adding styles to DESC

[Werner LEMBERG]

> [...]  why are we comparing troff to LaTeX? The original troff had
> no family selection request and you've demonstrated that the LaTeX
> font scheme is significantly different from groff's.

Because IMHO LaTeX's font selection scheme is far superior to what
groff has built in.  Note that I don't say that it is simple or well
suited for everything.

> If in traditional troff the canonical method for changing font is
> .ft, then why was .fam added to do the same thing?  It seems to me
> that .fam should be in the same class of request as .fp, .fspecial,
> and .ftr, which describe how font changes will occur the next time a
> .ft is called.  Isn't this an easier paradigm to learn?

Maybe, but this isn't how James Clark has implemented it.  There are
two categories of font commands:

  . Commands to register (`mount') fonts: .fp, .fspecial, .ftr, .sty,
    .special, .fspecial.  Obviously, these commands are `delayed'
    since the current font isn't changed.

  . Commands to select a font: .ft, .fam. Both commands act
    immediately.

> It's certainly not the only place where the notion of "delay"
> exists, e.g., requests beginning with the no-break control
> character delays use of particular values until the next line
> ('vs, 'in, etc.), which I think is philosophically related if not
> implemented the same way internally.

But font changes are *never* delayed.

> So, is it worth the effort to change the current behaviour?  At
> least, I think, it would make various fixes to various macro
> packages unnecessary.

Sorry, but I won't invest time to change that (as I've already stated,
I don't think that it should be changed at all).

In general, I believe that we abuse .fam more or less.  As it has been
implemented, in combination with the `-f' command line flag, its
intended use is to change the font family once, namely at the start of
a document.


    Werner