[TUHS] PWB contributions

Larry McVoy lm at mcvoy.com
Mon Nov 9 14:44:08 AEST 2015

On Sun, Nov 08, 2015 at 08:39:12PM -0500, Doug McIlroy wrote:
> >> I thought PWB (makers of "make") came from Harvard?
> > PWB ...  came straight out of Bell. Not sure about all the
> > applications (well, SCCS came from Bell).
> PWB did not create make; Stu Feldman did it in research.
> PWB did make SCCS. 

Marc Rochkind did SCCS, right?  Wikipedia said he did it at Bell Labs.

Does anyone on the list remember him?  I'd love to hear some stories
about how SCCS was created.

Warning: source management rant coming.

SCCS got a bad rap from Walter Tichy when he created RCS, I can't remember
his dislike but it's completely wrong.  RCS is diff based, you have a file
and a series of patches.  For the tip of the trunk, that's the file, going
backwards in history they are reverse patches.  OK, not so crazy.  But now
think about a branch.  You do reverse patches back to the branch point and
forward patches down to the tip of the branch.  Anyone living on a branch
hated RCS.

SCCS has interleaved deltas.   It's a brilliant design that has far far
better performance than anything else out there.  I'll explain that if
anyone cares but I'd love to know if anyone talked to Marc while he was
building SCCS.  Love to learn how he came up with that.

The reason I know so much about SCCS is that BitKeeper is sort of SCCS 
on steroids, we took the file format, fixed a bunch of stuff, and made
the first distributed version control system with it.  Git, Mercurial,
all the others, are copies of what we did.

I wish Marc was on this list, be fun to chat.


More information about the TUHS mailing list