One issue that came up at *every* client I ever worked for was version control for PeopleSoft applications. Chris Heller at Grey Sparling Solutions just did a very interesting post about this.

One of the things Chris mentions - although he doesn’t go into deep detail - is to note that PeopleSoft had a metadata-based architecture. Every PeopleSoft page - full of fields, text, everything - was NOT defined by a bunch of code. Instead, that entire page definition (as well as records, fields, and dozens of other types of objects) were all just stored as definitions in hundreds of database tables. Think of it as storing NIB files in a database - as rows in tables instead of as text fields. This always made version control a challenge - PeopleSoft was not a bunch of textual code files. Each “object” was spread across hundreds of rows in dozens of tables. But the cool part? Chris says that PeopleSoft’s actual codebase was about 10% the size of SAPs. Awesome.

In some future project, if I had to choose between challenging version control, or an architecture that allowed me to provide similar (better? functionality in one-tenth the code, I know what I’d choose.