Differences between LPMOO and LambdaMOO

• DGD is a disk-based server, and so LPMOO automatically benefits from the ability of DGD to swap infrequently used objects to disk. In the current implementation, MOO verbs are also objects.

• LPMOO utilizes DGD's binary state dumps to maintain database persistence. The binary files end up being much larger than the equivalent LambdaMOO db file, but the time required to checkpoint and read the database is significantly reduced. DGD does not fork while writing the state dump.

• LPMOO is written entirely in an interpreted language, LPC. This means it will probably never be able to compete with the LambdaMOO server in terms of speed. However, there are a few ways LPMOO attempts to make up for this: The core elements of the simulation are intended to be compiled into C and linked directly with DGD using the LPC->C precompiler that comes with DGD. Also, LPMOO itself compiles MOO verb code into LPC. This means that there is still essentially only one interpreter (DGD) running the simulation, not two.

• Two additional datatypes planned for a future version of MOO have already been added: floats and tables. Some extra built-in functions have been added to support the new datatypes.

• Support for open_network_connection() is currently only experimental. A separate DGD package is available to add networking support, which otherwise is not native to DGD. However, until a stable version of the package is released, its use is not recommended.

rob@ccs.neu.edu