Subject: Re: UNIX manpage history: CTSS RUNOFF From: Jerry Saltzer Date: 23/10/2011 18:04 To: Kristaps Dzonsons On Oct 23, 2011, at 9:09 AM, Kristaps Dzonsons wrote: > > On a side note: Pierce (CTSS source listing archiver) gave me > > permission to mirror your source, but as he's not sure who can > > assert copyright, I feel I should ask you as well. Do I have your > > permission to maintain a copy and directly link to your RUNOFF > > sources and manual? Kristaps, You have my permission. I believe that the only other entity that could assert copyright is MIT, and its claim would be weak, because I never assigned copyright on either the sources or the documentation. >> >> ...I was unaware that Bob was involved in any of the RUNOFF or >> >> roff ports, and I was surprised to see roff mentioned in his >> >> obituary. Your suggestion that he might have ported it to GE-635 >> >> assembler language seems quite far-fetched to me. It might be a >> >> good idea to contact Doug to get that part of the story straight. > > > > This is still a mystery, and I'm afraid it may stay that way without > > a source record. Everybody asserts that Bob Morris did something > > with RUNOFF and that it probably involved the GE-635, but nobody is > > quite sure what. Doug can only guess at the year, "I believe it > > happened after his year in Berkeley. That would probably make it > > about 1967 or 1968." Secondary sources erroneously note that > > Ritchie based a pre-UNIX PDP-7 version called rf on Morris' code, > > but it was in fact Thompson (by his own account), and based on > > McIlroy's BCPL re-write. A comment on the name "rf": Because the modems and the mechanical terminals of the day were slow, Multics gave most commands two names, a long descriptive name and a short one for ease of typing. Thus Multics "runoff" was also named "rf". (See the Multics Programmer's Guide for documentation.) When Thompson and Ritchie went off to develop Unix, they generally adopted just the short names. Thus the Multics "list | ls" became Unix "ls", etc. >> >> Quite separately, someone at Dartmouth College did a port of >> >> RUNOFF to DTSS... Given the rumored GE-635 connection, I wonder if Bob Morris might have had something to do with the Dartmouth College DTSS port? It may be worth looking for a DTSS historian and for the sources of the DTSS version of RUNOFF. I also recall that there was a GECOS-II TSS version of RUNOFF which might have its roots in something that Bob did. If Bob was involved, probably his first step would have been to port BCPL to GECOS II, rather than rewriting RUNOFF in 635 assembler. And it may be that the DTSS RUNOFF was just the binary of the GECOS-II TSS RUNOFF. >> >> The main question of this mail regards "compose", actually, which >> >> is another evolutionary path and one with some holes. "compose" >> >> was apparently a PL/I re-write of RUNOFF on Multics. I found the >> >> source for this utility, but it's poorly documented. Tom van >> >> Vleck asserts that Edwin Wallman had been working on this utility >> >> since around 1976 to eliminate dependence upon BCPL. This implies >> >> that it's based on Doug's BCPL runoff. However, while Ed is noted >> >> in the sources, the secondary record shows Dennis Capps as >> >> starting compose in 1974. Do you have any recollection of this >> >> branch? As you probably have figured out, Multics runoff was written in BCPL. I don't know who did that port, but it was done by recompiling Doug's CTSS ROFF using the version of BCPL that produced GE-645 code. The Multics BCPL sources of runoff from many years later are on-line, and they have Bob Mabee's name all over them, but he may have just been a later maintainer. I only vaguely recall hearing about the compose project--I don't recall any direct involvement. Tom Van Vleck is almost certainly the most authoritative source on that. > > Best wishes from a sunny Stockholm, And best regards from sunny Boise, Idaho. I'm glad to see someone working to get to the bottom of some of these historical mysteries. Jerry