CMU Andrew system (and wm) preservation

Michael Kerpan mjkerpan at
Mon Sep 7 14:29:51 CDT 2020

Has anybody even been able to get the X-based version to build? I remember
finding it on some Unix/Linux source code CD-ROM like 20 years ago,
thinking it sounded useful and cool, and trying to build it on whatever
Linux I was using on my hand-me-down 486 back in 1999/2000. Even back then,
it didn't build for me.


On Mon, Sep 7, 2020, 2:44 PM Chris Hanson via cctalk <cctalk at>

> On Sep 7, 2020, at 6:24 AM, dstalk at wrote:
> >
> > The description I have for AUIS (6.3.1) is:
> >
> > "AUIS (Andrew User Interface System) - compound document
> > environment offering a word processor, mail/bulletin board
> > reader/writer, drawing editor, spreadsheet, font editor,
> > application builder, and many other facilities"
> >
> > Again, an application, not a windowing system per se.
> Yes, the Andrew environment implemented proper layering, so ATK was made
> to work atop X and the applications (messages, ez, console, typescript,
> etc.) came along.
> At Carnegie Mellon in the early 1990s, you could (with only a little work,
> to use a console rather than graphical login) use either X or wm on some of
> the campus workstations. On a DECstation 3100 running Ultrix, if you
> weren’t going to run any X applications wm was *much* more responsive. I
> wasn’t around when the clusters had Sun-3 or IBM RT hardware but I can
> imagine the differences there were even more pronounced. (With wm, a
> DECstation felt as much faster than a Mac II as it actually was…)
> Applications built against ATK could run atop either wm or X; I don’t know
> if there were distinct builds of ATK or if the conditional logic was in the
> framework itself, but the applications themselves worked just fine with
> either since Andrew implemented a shared library mechanism. (Yes, even on
> Ultrix.)
> The publicly-released Andrew distributions don’t include the wm code, only
> the X version. I don’t know if they’ll actually build against the wm
> headers and libraries if they’re present, or if by the time CMU was
> releasing them publicly they had stripped that code out entirely.
>   — Chris

More information about the cctech mailing list