Subject: Re: UNIX manpage history From: Nils-Peter Nelson Date: Mon, 24 Oct 2011 06:12:02 -0700 To: Kristaps Dzonsons Runoff produced fixed width characters for a line printer or teletype. The Bell Labs version was called roff and served until linotype devices were available. The first troff port to the GE-635 was done by Roger Faulkner, now at Sun. though it may be that Bob Morris started it. I was indeed the supervisor of the DWB project, but I knew Kernighan well and the DWB version was always the same as his. However, that was inside A&T. Unix Systems Lab continued to sell a version of DWB with an old version; they never kept up. Then they dropped it completely, selling the rights to a company in Canada, SoftQuad. Eventually, USL started selling our version. [redacted] Ossanna'a original troff only worked with the typesetter in the Unix room. Whole idea of ditroff was to make it easier to put on back ends for other devices. We bought the first laser printer made (ours was the first on the East Coast), an Imagen Imprint 10, so that was an early post-processor. My recollection is that most of Kernighan's ditroff versions were bug fixes rather than language changes, though I think he did add features to make pic and eqn work properly. Rich Drechsler of my group wrote dpost, the postprocessor for PostScript devices, and from then on we only bought devices that used PostScript. HP must have had a dozen internal languages and we didn't want to be bothered. ----- Original Message ---- From: Kristaps Dzonsons To: Nils-Peter Nelson Sent: Mon, October 24, 2011 7:01:51 AM Subject: UNIX manpage history Nils-peter, Brian Kernighan suggested that I write to you on a matter of history, namely that of UNIX manpages. I've been constructing a primary-source account of the manpage from Saltzer's RUNOFF onward, inspired by the significant inconsistency of the secondary record. I think the "prehistorical" age is fleshed out (the sixties), with the exception of a mysterious GE-635 port by Bob Morris; the "classical" age (seventies) focussing on UNIX and proto-UNIX (Thompson, Ossanna, Ritchie, and Kernighan); and the open source "renaissance" (eighties and nineties) reimplementations of groff, awf, and cawf. And of course my own which neatly slips in as "next-generation" (exporting an AST). Kernighan suggested contacting you, I suppose (he didn't say exactly why), for an account of the continued branching of ditroff within AT&T. Unfortunately, I assume a lot of this code is still closed-source, and this makes it difficult to construct a source-record per se (are you aware of any primary sources for creating a primary record?). My current assumption is that Kernighan's device-independent troff continued its way through most AT&T projects (including the DWB), but I'm unsure of how much it changed over time. My interest, in this regard, is more in full ports/re-writes than repackaging unless it's a big event, like Heirloom troff after Sun released their troff. (I see your "Where does one get real troff these days?" thread; can you verify this is your work? If so, I can at least mention the various branches from Kernighan's troff.) I appreciate any time you can invest in answering, and I'm sorry I can't ask any more specific questions beyond asking for a general history and primary source references. Thanks, Kristaps