Re: groff/-mandoc replacement

[Thor Lancelot Simon <tls%rek.tjls.com@localhost>]

On Mon, Mar 09, 2009 at 10:48:26PM +0100, Joerg Sonnenberger wrote:
> On Mon, Mar 02, 2009 at 05:51:23PM -0500, Greg A. Woods wrote:
> > I also think it's ludicrous to consider trying to replace groff with
> > something half-backed that only barely attempts to handle just one macro
> > package's worth of documentation.
> 
> There is a very good reason to do that. mdocterm is over an order of
> magnitude faster than groff. Given that it is hard to benchmark on my
> laptop, some numbers from Hans on the VAX:

Holy crap.

> The various ditroff versions are somewhere around three times faster
> than groff, which still makes an extreme difference.

Have we ascertained that there's a free and portable ditroff yet?  It
would be regrettable to revert to a C/A/T troff -- though I bet it's even
faster.

The huge performance difference would, itself, seem to me to be a
persuasive reason to go to the trouble of adjusting the system to use
mdoc{term,ml}for HTML or ASCII output from -mdoc output, and ditroff
for everything else.  There's no mdocps, right?

Thor