Borland's Caliber RM

Just was given a brief presentation on Borland's Caliber RM (Requirements Management). It really fills the void that I saw in development methods I've used here. What I normally see, is requirements stored in Word documents, these documents are typically like a template with the same part in them. They are always out of date, and difficult to see what has changed over time. Caliber removes the need for Word, you store the requirements in a hierarchy and it stores everything in a database (they call it an OO repository). You can register who is a stakeholder and they will automatically receive e-mails when something changes in the requirement. Use cases, textual unit test descriptions, supporting documents can also be stored there. In addition you have some flexibility in that certain types of requirement can have specialized forms that need to be filled out.
The tool is actually quite simple, in many ways. But there are some minor and major issues with the system as well. One issue is that it doesn't use any standards like OPML to store it's information. The tree isn't versioned so moving things around doesn't get stored as a change. Deleting an item is permanent and all the related history of changes, etc. gets permanently deleted.
I had a similar idea to do the same stuff using ReStructured Text and a Subversion repository. One of the important things that Caliber would have on top of my solution is the ability to notify people of changes (although you could develop that for Subversion).
Caliber also has a nice glossary feature which automatically highlights words that are in the glossary. This encourages users to use the same terminology throughout the project. Also, words marked in the glossary as ambiguous are shown in red and a mouseover shows some help. For example, the words User Friendly is ambigious and should be replaced with something more specific.
Actually, now that I think of it all this can be done in my implementation, with just a little code to glue it all together seamlessly. Perhaps I'll file this away as another million dollar idea to do in the future.

Comments

Popular posts from this blog

Shortest Sudoku solver in Python

Dot to Png

Seven Segment Display in Inkscape