Re: [groff] Design and Implementation of *roff

[John Gardner]

Insanity. Seriously.

Reimplementing Troff is a task best not attempted at all. The grammar is
horribly complex, the language riddled with cryptic, arcane features, and
most importantly, it's not one program. It's several: the Troff pipeline
involves preprocessors, postprocessors, and bundled macro packages, each of
which have their own nuances and complications. You'd also be tasked with
handling font metrics for device-specific output (PDF, terminal,
PostScript, et al)...

Normally I don't discourage people from trying their own implementations,
but there are some things you just don't try: hand-rolling your own web
browser, writing your own compiler infrastructure, or reimplementing
complex, ancient typesetting systems like TeX or Troff.

You need to be a special sort of crazy (and patient, and knowledgeable) to
want to endure the obvious pains of reimplementing such a complex system.

Here's the source code <https://github.com/Alhadis/otroff> for the original
AT&T troff, BTW.

— John


On Sat, 1 Dec 2018 at 02:37, Yves Cloutier <address@hidden> wrote:

> Hello,
>
> I'm in search of any documentation that provides insight into the design
> and implementation of a *roff.
>
> I know source code exists for a few implementations like Plan 9, Heirloom,
> Groff and Neatroff.
>
> However what I'm in search of is something a bit more high-level than
> source code. Rather some design/implementation documentation that explains
> an *approach* to implementing a *roff.
>
> Would anyone know if anything like this exists?
>
> My google searches how not come up with anything, and was just wondering if
> someone somewhere might have such documentation lying around in their
> archives.
>
> regards,
>
> yc
>