Memory Model [all high priority]
(9)a Design and implement a runtime repository to maintain parsed data
(9)b Implement the client for accessing and using parsed data 

Memory

The underlying memory metric is that of "memory working set". This is the
minimum amount of memory required to run the application while achieving
adequate performance. Detailed goals for what the required memory should be were
not discussed. The team agreed that the selected solution should scale "nicely"
with application size, so for example, even large applications like Firefox
could run in standard system configurations. Additional quantification is needed
here.

The development team proposed adding a repository to the language model. Two
options were discussed: (i) use of Lucene, (ii) custom solution.

Development agreed to: (i) execute a due diligence effort to identify available
open source solutions (beyond Lucene) for a repository implementation, (ii)
simulate the memory and performance profile for Lucene with large production
applications, (iii) size the effort for a tuned custom implementation.

The selected repository approach needs to serve all language model needs,
including the symbol table (required for the accuracy work), the core parsing,
and a future cross reference ("xref"). The decision for which approach to use is
pending completion of exploring the listed options, and a top level design.