Garance Alistair Drosehn

Days leading up to MTS at RPI

Garance started out as a student at RPI in 1975 (and he actually arrived at RPI with the name Gary Robert Drosehn). At the time, RPI had a 360/67 mainframe which was extremely overloaded, and running OS/MVT (for batch-job processing) and Alpha (time-sharing). Almost all computing was done on OS/MVT, which is to say: using punch-cards.

RPI had a student chapter of the ACM on campus, which was a very large club. They had a public meeting with the Provost of RPI at the time, and "impressed" upon him that the students were really really unhappy with the state of computing. Garance was one of the students who spoke up at that meeting. After the meeting Garance was one of the students that met with Jim Moss, who as the Director of Computing Services at the time. Dr Moss was more than happy to have the students demanding better computer resources.

That meeting, combined with other events, convinced the administration that they needed to improve computer facilities. This included shopping around for a new operating system, and there was a committee to evaluate various candidates. Garance was one of two student representatives on that committee. I think the process was lead by either Don Porter or Wilson Dillaway. The committee selected MTS as the new operating system to use. By the fall semester of 1976, MTS was running on the mainframe for the first part of each day, and then later in the day it would switch back to OS/MVT.

Days officially working on MTS

In fall 1977 or spring, 1978 Garance audited a graduate-level systems programming course taught by Wilson Dillaway. Another student in the same class was Brian Eliot. In the summer of 1978 Garance was going to leave RPI, but Wilson offered both Garance and Brian jobs as "student systems programmers". Garance initially worked on the operator's job program, and the 3270 DSR.

Garance went on to have a one-year stint as a "Junior systems programmer", and then was hired as an official "systems programmer". He tended to work on the MTS job program, the 3270 and 3420 (tape) DSR, and several CLS's. This included things like PMF (from Jim Hansen), $MAKE, and the software management macros. He was also very interested in pulling in developments from other MTS sites to MTS at RPI, and he "made a name for himself" by doing that. RPI provided all of UofM's documentation to users at RPI, and thus it was often important that RPI pull in changes from UofM faster than was possible by using the official distributions.

He was keen on programs which helped collaboration between the MTS sites, such as MTS:FORUM (which used CRLT:CONFER at UofM), and *FORUM (written by Alan Ballard at UBC).

Garance was not the initial developer for much of anything on MTS (or at least, nothing that he remembers!), but did write major updates for some parts of the MTS job program, $MAKE, $PEEK, and SEG2:S2L. He also wrote minor updates to just about everything in the resident system, and most of the CLS's (Command Language Subsystems).

Over a few years in the early 1980's, "Gary Robert" changed his name to "Garance Alistair", and in fact that name change started because there were too many programmers with a first name of "Gary" in the MTS community! See http://www.rpi.edu/~drosehn/Personal/Gad-Name.html for the gory details.

Post-MTS work

In the late 1980's, RPI was moving away from mainframe computing, and towards unix workstations. Garance also bought a Mac Plus of his own, and found that the new Unix empire provided no reliable way for him to print. So, he did some work to provide printing for Mac users, and eventually took over the printing empire at RPI. "The printing empire" is mainly very customized versions of CAP and a version of 'lpr/lpd' which came from one of the BSD operating systems. Later that also included support for SAMBA servers for the RPI community. In the early and mid 1990's Garance provided technical support for NeXTSTEP, when the Campus Computing Store was selling NeXTstations to people at RPI, and at some nearby colleges.

RPI stopped active development of MTS by 1995, keeping it running mainly for some administrative applications that used MTS. MTS was turned off for good at RPI in summer of 1999. Basically, we (RPI) were scared of what we would do if MTS had any Y2K bugs in it...

Garance remains at RPI, mainly working on printing support, and general unix support. By 2005, that includes Redhat Linux support. Garance still does some Macintosh support, for some people running MacOS 10 at RPI. Garance is a committer in the FreeBSD project (at http://www.FreeBSD.org/), and also contributes to the OpenBSD (see http://www.OpenBSD.org/) and OpenAFS (at http://www.OpenAFS.org/) projects (mainly contributing money to those last two projects, not code...). Garance also provides the hardware and some support for a chat service called 'lily' (see http://www.freegroups.org/lily-twiki/bin/view/Lily/WebHome). This chat service is used mainly by RPI alumni, and friends of RPI alumni. This 'lily' chat server is actually a descendent from *FORUM on MTS. Not that it has any code in common, but RPI students keep writing new CMC's, and that started with their experience with *FORUM.

When he's not programming in C, Garance is using Ruby (see http://www.ruby-lang.org/en/) to write scripts, or is busy learning subversion (see http://subversion.tigris.org/). Subversion is an alternative to CVS, and some RPI graduates have worked on the project.

Things Garance still misses from MTS...

Going back to his student days, Garance was always interested in computing languages. MTS had the systems-programming language called *Plus, which was developed mainly by Alan Ballard and Paul Whaley at UBC. Garance liked *Plus when he first saw it, and all these years later he still wishes he could be programming in *Plus instead of C...

Garance is also still interested in the Command-Language-Parser (aka CLParser) which was written by Alan Ballard at UBC. Alan rewrote that in C, and Garance has a copy of that he hopes to use for some personal projects on Unix.

The MTS file-permissions model had features which are still not matched in most other environments, particularly in the area of PKEYs ("program keys"). Garance's interest in OpenAFS is partially because it has a much more flexible permissions-capability than Unix has, but OpenAFS doesn't have PKeys.

And Garance still thinks that *CDUpdate and *Compare in MTS did a much better job than 'patch' and 'diff' do in the Unix world.

(return to the MTS Wiki Home Page)