Notes:
I jettisoned MKS in favour of CVS and wrote up the working practices to be followed on the project. (This coincided with an ACCU-general discussion on version control - I posted a link to the write-up for review and got some very helpful input.)
Everything needed to build the system was placed under version control (well almost – the Websphere installation was too big).
The build scripts that Mark Goodwin had written were cumbersome because of the way in which that project's sourcecode was organised. Based on his experience I changed the organisation of code to simplify the build process, and wrote Ant build scripts that would build the system, run unit and smoke tests and build Javadoc.
As the project progressed it ramped up from the original three members to a dozen as programmers were added. The first thing developers noticed on joining the project was that they didn't spend the first couple of days (or more) struggling to set up a development environment. A developer joining the project could install Websphere, check out the "module" and immediately run a script that would build the system, smoke test it and construct the javadocs.
I introduced review meetings to:
Validate the design of packages (mostly sequence and class diagrams – although the package and class descriptions also figured); and
Validate the design of classes (Javadocs)
As these working practices were introduced they were incoporated into an introductory document provided to developers joining the project. (Based on my original draft this was maintained by another Mark – Mark Cooper.)