Cloverleaf started as a bet that it was impossible to easily make Clover into a multi-window creature. It evolved into a minor nightmare that taught the authors many lessons that they will hopefully learn the next time they do this. It was broken down into two main components, the core and zero or more leaves.

The core was really just a modified clover text client. Any messages that were not sent elsewhere would end up displayed on the core. The core was responsible for translating clover's bizzare network protocol and determining which leaves, if any, should get the message. It was also responsible for listening for new leaves and handling their requests.

Each leaf was an independent program that communicated with the core through a named pipe (or was it a UNIX domain socket?). It would send messages to the core, and would receive already formatted messages from the core that it would display. The leaves were relatively simple programs, but had some special features. Each leaf could play a different audible signal for those events that had signal enabled, so you had audible cues for which windows were active. Each leaf also had its own default sendlist which usually matched all the users and discussions in that leaf, but which you could override using the typical destination;message format.

One notable feature was that all of the programs were text-based. They could be run inside 'screen' sessions or in separate xterms, whatever the user wanted. Several leaves were available:

• Clover-leaf was a Clover-like interface that isolated one user or discussion in their own window.
• Talk-leaf was a talk-like interface.
• Bridge was a demonstration that Cloverleaf could be used to send messages between two Clover servers. It was later used, almost without modification, as part of the Clover-lily bridge.
• Lumberjack was a demonstration that Cloverleaf could be used to record transcripts of discussions.

Historical note: The "leaf" option in lily is named after CloverLeaf.
to top