X-Git-Url: http://lambda.jimpryor.net/git/gitweb.cgi?a=blobdiff_plain;f=lambda_evaluator.mdwn;h=d39086fd257131bb012c1fa5f71191c387c23927;hb=51f881475438eba8f8533e5e0a91ee087e72aaaf;hp=31021dcee2413514cbbec8803bd51c52e0f947d4;hpb=904d62aef970037420a9f3e78d9ebc994e524063;p=lambda.git
diff --git a/lambda_evaluator.mdwn b/lambda_evaluator.mdwn
index 31021dce..d39086fd 100644
--- a/lambda_evaluator.mdwn
+++ b/lambda_evaluator.mdwn
@@ -1,12 +1,9 @@
-Lambda Evaluator
-----------------
-
-There is now a [lambda evaluator](http://lambda.jimpryor.net/lambda-let.html) available.
-It will allow you to write lambda terms and evaluate them, with full ability to inspect the results.
+This lambda evaluator will allow you to write lambda terms and evaluate (that is, normalize) them, and inspect the results.
(This won't work in Racket, because Racket doesn't even try to represent the internal structure of a function in a human-readable way.)
*Lambda terms*: lambda terms are written with a backslash, thus: `((\x (\y x)) z)`.
-If you click "Reduce", the system will produce a lambda term that is guaranteed to be reduction equivalent (`<~~>`) with the original term. So `((\x (\y x)) z)` reduces to (a lambda term equivalent to) `(\y z)`.
+
+If you click "Normalize", the system will try to produce a normal-form lambda expression that your original term reduces to (~~>). So `((\x (\y x)) z)` reduces to `(\y z)`.
*Let*: in order to make building a more elaborate set of terms easier, it is possible to define values using `let`.
In this toy system, `let`s should only be used at the beginning of a file. If we have, for intance,
@@ -15,23 +12,133 @@ In this toy system, `let`s should only be used at the beginning of a file. If w
let false = (\x (\y y)) in
((true yes) no)
-the result is `yes`. Things to watch out for: the expression after the equal sign must have balanced parentheses,
-and the "in" is obligatory. If you violate these rules, the system will still produce a result, but it won't make much sense.
-
-*Abbreviations*: No abbreviations work. So `\xy.yxx` must be written `(\x (\y ((y x) x)))`.
+the result is `yes`.
*Comments*: anything following a semicolon to the end of the line is ignored.
Blank lines are fine.
+*Abbreviations*: In an earlier version, you couldn't use abbreviations. `\x y. y x x` had to be written `(\x (\y ((y x) x)))`. We've upgraded the parser though, so now it should be able to understand any lambda term that you can.
+
+*Constants*: The combinators `S`, `K`, `I`, `C`, `B`, `W`, `T`, `M` (aka ω
) and `L` are pre-defined to their standard values. Also, integers will automatically be converted to Church numerals. (`0` is `\s z. z`, `1` is `\s z. s z`, and so on.)
+
+*Variables*: Variables must start with a letter and can continue with any sequence of letters, numbers, `_`, `-`, or `/`. They may optionally end with `?` or `!`. When the evaluator does alpha-conversion, it may change `x` into `x'` or `x''` and so on. But you should not attempt to use primed variable names yourself.
+
+
+
+
+do eta-reductions too
+
+
+
+
+
+
++ + + + Under the hood --------------- -The interpreter is written in JavaScript (which is not closely related to Java), and runs inside your browser. +The interpreter is written in JavaScript and runs inside your browser. So if you decide to reduce a term that does not terminate (such as `((\x (x x)) (\x (x x)))`), it will be your browser that stops responding, not the wiki server. -You can inspect the code [here](http://lambda.jimpryor.net/code/lambda.js). Suggestions for improvements welcome. +The main code is [here](http://lambda.jimpryor.net/code/lambda.js). Suggestions for improvements welcome. + +The code is based on: + +* Chris Barker's JavaScript lambda calculator +* [Oleg Kiselyov's Haskell lambda calculator](http://okmij.org/ftp/Computation/lambda-calc.html#lambda-calculator-haskell). +* The top-down JavaScript lexer and parser at