Re: [Groff] First round of cleabnup patches

[Eric S. Raymond]

Werner LEMBERG <address@hidden>:
> My fault, wrong question.  Another try: What's the problem?  The
> current code I have is this:
> 
>   .if n \{\
>   .  do ftr CR R
>   .  do ftr CI I
>   .  do ftr CB B
>   .\}
>   .
>   .if '\(.T'dvi' \
>   .  do ftr CB CW
> 
> groff's two-letter `do' instruction nicely hides extended groff
> requests...

That's not the problem I see.  The problem I see is that this code wires in a
set of assumptions about font mappings that won't necessarily be wanted by
everyone who wants to use .SY/.OP/.YS.  .SY/.OP/.YS are mechanism; this is
policy.  

I generally find it a good idea to keep mechanism separated from
policy, if only because when I don't I get developers annoyed at me
because they have tripped over policy choices they did not know they
were importing along with the mechanism.

> > I'm now a bit puzzled about why an-old and an-ext have separate
> > existences at all.  If we're going to dicument the an-ext things and
> > treat them as first-class citizens, why have two files?
> 
> The idea was to have a set of macros (written in pure classical troff)
> which is (a) in the public domain, (b) which runs with all troff
> incarnations, and (c) which man writers might use as a template for
> copying the relevant snippets to the man page's preamble.
> 
> I thought we've agreed on that.

I had to puzzle over your "I thought we've agreed on that" for a while
before I got what you were driving at.  We sort of did agree, but my
mind was on other aspects of several related problems and I didn't 
realize that the split between an-old and an-ext was part of
a plan for propagating these extensions to other *roffs.

Now that my attention has been focused on it, that part makes sense.
I withdraw my non-objection :-).

As soon as you tell me you've finished the merge, I'll resync, look
at what you've dome, and start working on (a) more cleanup, and (b)
documentation.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>