Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!julius.cs.uiuc.edu!ux1.cso.uiuc.edu!ux1.cso.uiuc.edu!m.cs.uiuc.edu!johnson
From: johnson@m.cs.uiuc.edu
Newsgroups: comp.lang.smalltalk
Subject: Re: PPS overpriced = low marketshare
Message-ID: <5600012@m.cs.uiuc.edu>
Date: 3 Oct 90 18:42:00 GMT
References: <171937@<1990Sep28>
Lines: 30
Nf-ID: #R:<1990Sep28:171937:m.cs.uiuc.edu:5600012:000:1381
Nf-From: m.cs.uiuc.edu!johnson    Oct  3 13:42:00 1990


Smalltalk-80 works fine for multiprogrammer projects.  The
change management tools are an important part of its success,
since these tools let you know exactly which part of the
image have been changed and whether several sets of changes
will conflict.

There are currently four people working on the project, down
from six last summer.  The project is large, so there are
parts of it that are fairly independent, but we often have
several people working on the same classes.  As long as they
behave properly, this is not a problem.

Ideally, people file out their changes every day or so and
put them in a common directory.  The files are ordered by
the time they were created.  The rule is that a new file
must be compatible with all the ones before it.  Sometimes
a new file will not be compatible with changes that someone
is currently working on.  That is the fault of the person
who has NOT installed his changes.  This rarely happens
unless someone has been sitting on changes for weeks and weeks.
Most change files are small; a dozen methods or less.  Obviously,
major changes cause more problems.

I have heard a lot more complaints from people using Smalltalk-V,
since it does not have the same change management tools.  However,
Smalltalk-80 is better for multiprogrammer projects then Unix with
RCS and make.

Ralph Johnson -- University of Illinois at Urbana-Champaign