1 files changed, 65 insertions, 0 deletions
diff --git a/base/lang-process.md b/base/lang-process.md
new file mode 100644
index 0000000..1175fca
--- /dev/null
+++ b/base/lang-process.md
@@ -0,0 +1,65 @@
+# Top level reader loop
+First, a command is read in from the user under a prompt
+
+Next, it is checked if the first token of the command (split on " ")
+corresponds to an action name
+* If it does, the corresponding action is executed, and the next command is
+  read
+* If it doesn't, the command is parsed as a language command and executed
+
+This continues until the command "quit" (non case-sensitive) is read
+
+# Known actions
+The currently implemented actions are:
+* env: print out the contents of the enviroment. Takes no arguments
+* inline: inline variables out of the definition of specified variables. Takes
+  at least two arguments
+  * The first is the name of the variable to do inlining in
+  * The second is the name of the variable to bind the inlined expression to
+  * The third and following are the names of the variables to inline in the
+	variable you are inlining. If you don't give any, it will inline every
+	variable
+
+## Details on inlining
+The way the inlining process works is simple. The tree for the variable to be
+inlined is read, and then each variable reference is inlined if it is marked as
+one of the variable references to inline. This only occurs one layer deep
+
+# Parsing language commands
+Once it is decided to parse a command as a language command, it goes through
+four steps:
+* Preparation
+* AST Building
+* AST Transformation
+* AST Evaluation
+
+# Command preparation
+Command preparation means turning the raw space-seperated tokens into tokens
+suitable for feeding the AST builder. It involves the following steps:
+* The first is operator expansion, which turns a token like 4+4 into the three
+  tokens 4, +, and 4
+* The next is token deaffixation which turns a token like (4 into the tokens (
+  and 4. However, ((4 will become (( 4, not ( ( 4, because (( is a different
+  nesting level than (
+* Next, any blank tokens that have been created as a result of the process are
+  disposed of
+* Finally, the entire expression is shunted, turning it from infix notation to
+  postfix notation. The only real special thing about this is that (( is a
+  different nesting level than (. The token precedence levels are (sorted from
+  lower to higher precedence): 
+  * Math precedence
+   * + & -
+   * * & /
+   * 
+  * Dice precedence
+   * d
+   * c
+  * Expression precedence
+   * =>
+   * :=
+
+# AST Building
+AST building is a slightly more complex process, as the tree is traversed
+twice: once to build the tree, and then again to convert the tokens from
+strings into data tokens. The only really finicky part about the first tree is
+arrays, because arrays are still in infix form at this point.