David Wilson

Old TODO: 1. User-defined key blob types. Allocate a small range from the key encoding to logic that looks up a name for the byte from metadata, then looks up...

The current tests are a joke. Needs at least: - Full set of tests written for py-lmdb (py-lmdb #21) - Full set of tests for key encoder. - Full set...

acid.keylib.packs() currently knows nothing about inverting keys. Some hacks can be used to mostly avoid the necessity, e.g. negating numbers, xorring strings, etc., but this is unintuitive and error prone....

This has been discussed a little via e-mail, but it's too far in the future to think about, so marking unlikely. Mostly covers the following old TODOs: 1. Support "read-only"...

Old TODO entry: 1. Be generic enough to allow indices and constraints on purely in-memory collections, without encoding overhead. Keeping it since it's an interesting target. There may be no...

Calling .get_bytes() creates a temporary string. Instead the '.int' property is the most efficient to access and construct from, but there is no PyLong_\* to provide access to the internal...

Related to more general notion of mapping the keyspace more cleanly.. at some point it would be useful to be able to swap in replacement collections/indices, preferably having the result...

The current encoding throws away a byte for every integer encoded: since only 10 values of the discriminator byte are currently used. We could instead collapse the first byte of...

acid.encoders.RecordEncoder is now smart enough to know how to access attributes from the record. It is trivial from there to extend core.py to support index specifications like: `"name, -age, location"`,...

Refactor much of what is there into core.rst, rewrite intro to use meta. Walk through a complete example from demo/ directory. Discuss some core ideas more thoroughly