Re: [Axiom-math] (no subject)

[Martin Rubey]

I copy this to axiom-math, because others might be interested.

C. Frangos writes:
 > 
 > Dear Martin,
 > 
 > I have modified your function test() in two ways (see below). Both 
 > modifications result in errors.

I give the reasons below.

 > I have also attached an attempt at a simple package which does not compile.

dito.
 
 > I am using the basic syntax given in the axiom book for command line 
 > programming, that is, semicolons at end of lines, brackets for function 
 > definitions, brackets for blocks in if-then-else, for-repeat statements, 
 > brackets for function calls, etc. I find this easier to remember and 
 > consistent. I dont favour indentation.

 > It seems that this syntax cannot be used in .input and .spad files. As you 
 > mentioned axiom seems to be very indentation sensitive (first time I come 
 > across this in a compuetr language).

I don't like it either. Phyton does the same thing (probably got it from
Axiom). However, the rules are very simple:

One operation per line. No semicolon at the end.

A couple of operations indented by the same amount constitute a "pile". In C
and Aldor, they would be enclosed by { }.

A pile is needed after

*  ==, i.e. a function definition

* for ... repeat
* if ... then

* else

That's all, I think.

 > If this is right then these undesirable features alone makes it difficult to 
 > use axiom. The seemingly different syntax for command line programming 
 > versus 
 > .input and .spad files is not explained anywhere in the axiom book as far as 
 > I could see ???

Well, both forms are allowed in .spad and .input files. (op1; op2; ...; opn) is
simply a construct to fit n operations in a place where only one is allowed.

 > (In axiom is there a read command ala C/C++ to allow the user to type in 
 > something, etc ??)

I'm afraid no. The approach is different. I'll help when we come to it.

 > ------ File test.input
 > 
 > g: INT := 5
 > 
 > -- Your original code.
 > --test():INT == 
 > --  output(g)
 > --  g := 6
 > --  l: INT := 10
 > --  output(l)
 > --  l
 > 
 > 
 > -- Code below compiles but identation causes error when you
 > -- attempt to run test(); !!!! 

correct. 

 > test():INT == 
 > 
 >   output(g)
 > 
 >   g := 6
 > 
 >   l: INT := 10
 > 
 >    output(l)
 > 
 >    l

You can see what )re test.input did to your function definition by typing

test

(without parenthesis) at the prompt:

(16) -> test

   (16)
   test () ==
     output(g)
     g := 6
     l : INT  :=
       10
          output(l)
          l


The following would work.

test():INT == 
 
  output(g)
 
  g := 6
 
  l: INT := 10
 
  output(l)
 
  l

 > 
 > -- Code below does not compile.
 > --test():INT == (
 > --  output(g);
 > --  g := 6;
 > --  l: INT := 10;
 > --   output(l);
 > --   return(l) )

The basic rule in axiom is: one operation per line.
The (operation-1; operation-2; ...; operation-n) construct constitutes one 
single operation (the
result is the result of operation-n), therefore it has to go on one line. The
following would work:

test():INT == (_
  output(g);_
  g := 6;_
  l: INT := 10;_
   output(l);_
   return(l) )

-------------------------------------------------------------------------------

Only two simple indentation mistakes in the following:


 > -------File zmean.spad
 > 
 > F ==> Float
 > 
 > )abbrev package XZMEAN xzmean
 > xzmean(): Exports == Implementation where
 >     Exports == with
 >       mean: List F -> F
 >       stddev: List F -> F
 >     Implementation == add
 > 
 >       mean l == 
 >         n := #l
 >         if n = 0 then error "The mean of an empty list is not defined"
 >         reduce(_+, l)/n
 > 
 > -- The following code does not compile.
 >       stddev l == 
 >        n := #l
 >        if n = 0 then 
 >       ( error("The standard deviation of an empty list is not defined") )
         ^^^ 

should be 
 >         ( error("The standard deviation of an empty list is not defined") )

 >        m := mean l
 >        s := 0.0
 >         for i1 in 1..n repeat 
          ^^^
should be
 >        for i1 in 1..n repeat      

 >          ( s := s + (l(i1) - m)**2 )
 >        s := sqrt(s/n)






 
 > -- The following code also does not compile.
 > --      stddev(l) == (
 > --        n := #l;
 > --      if n = 0 then 
 > --        ( error("The standard deviation of an empty list is not defined") )
 > --        m := mean(l);
 > --        s := 0.0;
 > --        for i1 in 1..n repeat
 > --          (s := s + (l(i1) - m)**2)
 > --        s := sqrt(s/n);
 > --        return(s) )

same reason as in the test.input example: (op1; op2;...;opn) is a single
operation, therefore must go on a single line. Use _ ! Furthermore, you forgot
two semicolons:

      stddev(l) == (_
        n := #l; _
      if n = 0 then _
        ( error("The standard deviation of an empty list is not defined") ); _
        m := mean(l);_
        s := 0.0;_
        for i1 in 1..n repeat_
          (s := s + (l(i1) - m)**2); _
        s := sqrt(s/n);_
        return(s) )