v2.2.0

TODO

@@ -1,11 +1,13 @@
 Features
 --------
 
-complete XQuery Fulltext support
+Complete XQuery Fulltext support
 XQueryP support
-Grouping support
-XSLT 2.0 support
-Higher order functions
+XQuery 1.1 Grouping support
+Complete XSLT 2.0 support
+Support pre-parsed modules
+Improve exceptions heirachy, and implement XQuery 1.1 try/catch
+Implement EXSLT math functions
 
 
 Optimization
@@ -13,10 +15,12 @@ Optimization
 
 Hash map implementation of distinct-values()
 nested loop elimination (group recognition, join recognition)
-variable inlining
 common sub-expression extraction
 faster modifiable data model
+Document projection for recursive functions
 pull event execution
+Remove and free unused global variables
+Better static typing esp. for higher order functions
 
 
 Infrastructure
@@ -27,10 +31,11 @@ string pool that returns numeric identifier
 support for UTF-8 execution?
 extract basic schema types from Xerces-C
 abstract schema type access (no direct access to Xerces objects)
-Build without code that needs a source version of Xerces-C?
 Improve MAPM concurrency
 Windows -- re-enable warning 4996 (xqilla_config.h) and address the "unsafe" warnings
 	for use of strcat, strcpy, etc
+Write query plans to byte code, and execute the byte code in a virtual machine
+Save out parsed query to disk
 
 
 QA
@@ -41,4 +46,16 @@ fix XQTS bugs
 new test suite
 test XQuery Fulltext
 Fix ATDoubleOrDerivedImpl and ATFloatOrDerivedImpl to use double and float types
+Make xs:decimal infinite length
+Fix numeric output methods to use variable size buffers
+Fix module import to work correctly, including:
+  - using the correct static context
+  - returning known schema types
+  - checking import circularities correctly
 
+
+Query Prepare Optimization
+--------------------------
+
+Rewrite DatatypeLookup to be cheaper and more efficient - maybe write our own primitive type handling?
+DocumentCacheImpl is expensive to create, maybe don't always create a scanner? Maybe handle primitive schema types ourselves?