design.md

 # Preliminary Design
 
-(note: some of this is already outdated, will revise)
-
 ## Reading
 
 To do macros we need to `read`. To `read` we need to match delimiters
@@ -19,27 +17,73 @@ disambiguate. Algorithm:
     skip over comments
 
     if tok is /
-        if tok-1 is )
-            look back to matching (
-            if identifier before ( in "if" "while" "for" "with"
+        if tok-1 is ()
+            if tok-2 in "if" "while" "for" "with"
                 tok is start of regex literal
             else
                 tok is divide
-        if tok-1 is }
-            if end of function expression // described below
-                tok is start of divide
+        else if tok-1 is {}
+            if isBlock(tok-1)
+                // named or anonymous function
+                if tok-2 is () and tok-3 is "function" or tok-4 is "function"
+                    if function expression   // how to determine is described below         
+                        tok is divide
                     else                            
                         tok is start of regex literal
-            
-        if tok-1 in punctuator // e.g. ";", "==", ">", "/", "+", etc.
+                else
                     tok is start of regex literal
-            
-        if tok-1 in keywords // though some keywords will eventually result in a parse error
+            else
+                tok is divide
+        else if tok-1 in punctuator // e.g. ";", "==", ">", "/", "+", etc.
+            tok is start of regex literal
+        else if tok-1 in keywords 
+            // though some keywords will eventually result in a parse error (eg. debugger, break)
             tok is start of regex literal
-            
         else
             tok is divide
 
+
+    assignOps =  ["=", "+=", "-=", "*=", "/=", "%=", 
+                    "<<=", ">>=", ">>>=", "&=", "|=", "^=", ","];
+
+    binaryOps = ["+", "-", "*", "/", "%","<<", ">>", ">>>", 
+                    "&", "|", "^","&&", "||", "?", ":", 
+                    "instanceof"
+                    "===", "==", ">=", "<=", "<", ">", "!=", "!=="];
+
+    unaryOps = ["++", "--", "~", "!", "delete", "void", "typeof", "throw", "new"];
+
+    function isBlock(tok)
+        if tok-1 is ( or [
+            // ... ({...} ...)
+            return false
+        else if tok-1 is ":" and parent token is {}
+            // ... {a:{...} ...}
+            return isBlock(the parent {})
+        else if tok-1 is one of assignOps unaryOps binaryOps
+            // ... + {...}
+            // ... typeof {...}
+            return false
+        else if tok-1 is "return"
+            // handle ASI
+            if lineNumber(tok) isnt lineNumber(tok-1)
+                // return
+                // {...}
+                return true
+            else                                
+                // return {...}
+                return false
+        else if tok-1 is "yield"  // could also put "yield" in unaryOps
+            return false
+        else if tok-1 is one of "debugger" "break" "continue" "throw"
+            parse error
+        else if tok-1 is one of "void" "typeof" "in" "case" "delete"
+            // ... in {...}
+            return false
+        else                                    
+            return true
+
+
 Depending on context, `function name() {}` is either a function declaration or a
 function expression. If it's a function expression then
 a following `/` will be interpreted as a divide but if it's a
@@ -58,9 +102,9 @@ following imply it is a function declaration:
     ; } ) ] ident literal (including regex literal so need to be careful about /)
     debugger break continue else
 
-And these imply it is an function expression.
+And these imply it is a function expression.
 
-    ( { [ , (assignment operators) (binary operators) (unary operators)
+    ( [ , (assignment operators) (binary operators) (unary operators)
     in typeof instanceof new return case delete
     throw void
 
@@ -139,45 +183,6 @@ etc.) as really reserved. Macros can't override their meaning.
 Should we disallow FutureReservedWords too (`class`, `enum`, etc.)?
 
 
-## Scope
-
-Macro definitions should be scoped appropriately.
-What should we do about hoisting? For example,
-
-    macro foo { ... }
-    
-    function bar() {
-        foo(...)
-        
-        if (x) {
-            var foo = function () { ... }
-            foo(...)
-        } else {
-            var foo = function () { ... }
-        }
-    }
-
-Because of hoisting, the variable `foo` will shadow the macro definition of
-`foo` in this code. But this is complex and annoying. Can we just say
-that hoisting always happens after macro expansion? So the first
-`foo(...)` is a macro invocation and the second `foo(...)` is a
-function call? Does this cause any problems?
-
-So I think we have to have hoisting happen after macro expansion. But what does this mean for hygiene?
-
-    macro foo { ... }
-    
-    foo { bar = 4 }
-    
-    var bar = 5;
-
-## Hygiene
-
-Should be [fun](http://www.quotationspage.com/quote/26964.html)...
-
-## Modules
-
-Details about importing macros from another module...
 
 ## Example Code
 
@@ -445,25 +450,10 @@ These names aren't quite right. What is JavaScripty?
       }
     }
 
-## Misc
-
-  * sub-form expansion? MTWT says parse and expand must be separate to do sub-form expansion.
-
-## potential macros
-
-heredoc or multi line strings.
-
-    s = heredoc {this is a nice
-        multiline string
-            that keeps spaces
-        the only problem is it can't break tokenization...
-    }
-
-string templates
 
 ## Papers
 
-Papers I've been looking through:
+Useful papers on macros:
 
   * Macros that work
   * Macros that work together
@@ -476,6 +466,3 @@ Papers I've been looking through:
   * Fortifying Macros
   * Composable and Compilable Macros
 
-What others might be helpful?
-
-