[erlang-questions] erl2 tutorial

[Joe Armstrong]

Tutorials 1 and 2
=================

   Download from
   https://github.com/joearms/erl2

Tutorial 1
==========

Here is a 10 line erl2 program:

(see if you can guess what it does before reading the explanation)
(( and tell me why if you guessed wrong))


$ cat example1.erl2

   defMods mod1 shell end.
   addMod shell.
   def fac = fun(0) -> 1; (N) -> N*fac(N-1) end.
   120 = fac(5).
   F25 = fac(25).
   addMod mod1.
   def a(N) -> F25 + N end.
   addMod shell.
   io:format("mod1:a(10)=~p~n",[mod1:a(10)]).
   erl2:make_mods().
	
To run the program we say:

$./erl2 example1.erl2
 Created:"gen/exprs.tmp"
 mod1:a(10)=15511210043330985984000010
 make_mods saving generated code...
 Created:"gen/all.gen"
 Created:gen/mod1.erl
 Created:gen/shell.erl
 Success

this generates two files

 $ cat gen/shell.erl
  -module(shell).
  -compile(export_all).

 fac(0) ->
     1;
 fac(N) ->
     N * fac(N - 1).

 $cat gen/mod1.erl
 -module(mod1).
 -compile(export_all).

 a(N) ->
     F25 = 15511210043330985984000000,
     F25 + N.

What do the commands do?

1) Start by defining which modules we will be creating

>  defMods mod1 shell end.

2) addMod X - means "we are adding code to the module X" so

>   addMod shell

   means we are adding some code to the module shell

3) Define a function

>  def fac = fun(0) -> 1; (N) -> N*fac(N-1) end.

   Defines a function fac (in shell since we add adding code to shell)
   Note: we can write self-referential functions here.

4) test the code
>  120 = fac(5).

5) Run the code

> F25 = fac(25).

6) Now add some code to mod1

>   addMod mod1.
>   def a(N) -> F25 + N end.

7) Go back to adding code in the shell

> addMod shell.

8) Test the function works

> io:format("mod1:a(10)=~p~n",[mod1:a(10)]).

9) Generate the code

> erl2:make_mods().
	
That's it

Now we can look at the generated code

Look at the code in mod1 - The shell binding for F25 was propagated into
the shell.

$ cat gen/shell.erl
 -module(shell).
 -compile(export_all).

 fac(0) ->
     1;
 fac(N) ->
     N * fac(N - 1).

 $cat gen/mod1.erl
 -module(mod1).
 -compile(export_all).

 a(N) ->
     F25 = 15511210043330985984000000,
     F25 + N.

Tutorial 2
==========

Now we change the code to the following:

 $ cat example1_fail.erl2

   defMods mod1 shell end.
   addMod shell.
   def fac = fun(0) -> 1; (N) -> N*fac(N-1) end.
   100 = fac(5).
   F25 = fac(25).
   addMod mod1.
   def a(N) -> F25 + N end.
   addMod shell.
   io:format("mod1:a(10)=~p~n",[mod1:a(10)]).
   erl2:make_mods().
	
This is *identical* to example1.erl2
BUT the unit test will fail since fac(5) is NOT 100

Now we run the program:

$ ./erl2 example1_fail.erl2
  Created:"gen/exprs.tmp"
  Error:{{badmatch,120},[{erl_eval,expr,3,[]}]}
  Compile failed

Now NO CODE is generated.

So

  1) If the test fails the code is not compiled.
  2) The unit tests are NOT run after the code is compiled
     but *before*
  3) Compile modules failing the unit tests are *impossible* to create

Cheers

/Joe