Open items related to the CouchDB project that anyone can work on.
The documentation, especially the introductions, could use some basic diagrams that show various the parts of a CouchDB server: Host OS, Application HTTP server, CouchDB database server (Server contains database which contain document which contain computed tables etc).
Right now CouchDB uses the ICU library from IBM for collation support. CouchDB currently uses a rather slow method of interfacing with the library, which is to copy the UTF8 string lists to binary buffers and then provide the buffers to ICU to do the comparison. More optimized methods of interfacing with ICU could be used, or just do the collation in native erlang using generated code to create the collation tables.