Create ravi-new-parser-codegenerator.rst

readthedocs/ravi-new-parser-codegenerator.rst

@@ -0,0 +1,27 @@
+New Parser and Code Generator for Ravi
+======================================
+
+These are some design notes on the new parser / code generator.
+
+There will be several phases:
+
+1. Convert input into syntax tree - this will look very much like the input source, i.e. there will be fornum loops, if statements,
+   while and repeat loops etc. During this phase only the types for literals and local decalarations will be known.
+2. The next phase will be a type checking phase. In this phase types will be derive for expressions, and also type assertions will be
+   added where required such as unpon function entry, and after function calls. But the overall syntax tree will still resemble the input 
+   source except for the additional instructions.
+3. Third phase will be to assign virtual registers; first to locals, and then to temporaries. For simplicity I will probably keep the
+   temporaries and locals in separate ranges.
+4. Next phase will be linearize the instructions - during this phase the loops, if statements etc will get translated to equivalent of 
+   conditional and unconditional jumps. 
+5. We could at this stage generate byte code and finish. Initially maybe we will stop here.
+6. The next phase will be translate into basic blocks.
+7. Following that we will construct a CFG, perform dominator analysis and convert to SSA form.
+
+Some other things
+-----------------
+1. locals that are used as up-values or passed or overwritten by function calls should be marked as having 'escaped'.
+   Having this knowledge will enable backend JIT to use the stack for values that cannot escape.
+2. during code generation it will be good to know which registers are type constant - i.e. their types do not change. register allocation
+   should be designed / implemented so that we try to avoid over-writing type data where possible. This will allow backend JIT 
+   to generate more optimized code.