OpenEuphoria

Re: Structure Pre-Processor

• Posted by Pete Eberlein <xseal at HARBORSIDE.COM> Feb 18, 1999
• 442 views

Bernie wrote:
> PETE
>
> What are all of the characters at end of your e-mail on listserver ??
>
> I presume they are some sort of attachment. How Do you decode them ??
>
> I use EUDORA for my e-mail .

Sorry.  I forgot that Pine sometimes encodes attachments that get mangled
on the way through the listserver.  It was supposed to be type TEXT-PLAIN.
I've posted the file to my webpage if you still want it.

Ralf wrote:
> Interesting approach. This way we do have the speed benefit, that comes
> with structures.
> You say 'scope not enforced', but the scope is already enforced by
> Euphoria.

Yeah, I guess Euphoria does catch some of the errors, but spp still
thinks all structure variables are non-local to routines...

    somestruc a

    function b()
        anotherstruc a
        ....
    end function

    a.whatever -- the preprocessor assumes the type of "a" to be
               -- "anotherstruc", not "somestruc" as expected.  But it
               -- only displays useless warnings, and will still translate
               -- it to something usable.

> The only real 'issue' that I can find is that we are not able to express
> the whole structure or parts of it.
> But this can be solved as well, I think. I see three options:

Yeah, I'd like to be able to express a structure as a whole as well...

>    1) When we use the whole structure, just built it up:
>         { my_structure_item1, my_structure_item2 }
>       Note: WHen we assign something to it, we would need to use some
> dummy variable.
>       However, this is *slow*

something like:

    structure rgb integer red, green, blue end structure

    rgb color
    color = {100,150,200}

should translate to:

    color_red = 100
    color_green = 150
    color_blue = 200

or in the case of not having a bracketed sequence,

    color = somefunction()
to:
    junk = somefunction()
    color_red = junk[1]
    color_green = junk[2]
    color_blue = junk[3]

Reading a whole structure could be pretty easily translated:

    ? color
to:
    ? {color_red, color_green, color_blue}


>   2) Always keep the decleration of a variable (using a structure)
> global.  And have a little function to retrieve and set the structure.

This could get messy... I wouldn't want to try it.

>     3) Use at least one level of slicing. That is, structures in
> structures, etc. are automatically resolved, only the
> first-top-level of the structure is a sequence. Maybe a little database
> managed by an external library.

This sounds overly complex.

>         Faster than the above options. Esspecially since the slicing
> could be partly 'optimized' away, by efficiently lookup and
> store values in the sequence, only when needed.

Slicing a structure?  I don't get it.

>     Oh, and why cant we use structures in a routine ? I mean, without
> passing it around, we should be able to use them in a
> routine, or am I missing something here ?

What I mean is, you can have them in the arguments in a routine...

    structure test
        integer i,j,k
    end structure

    procedure funky(test monkey)
        test goober
        goober.i = monkey.i
        goober.j = monkey.j
        goober.k = monkey.k
    end procedure

will translate to:

   procedure funky(integer monkey_i, monkey_j, monkey_k )
       integer goober_i, goober_j, goober_k
       goober_i = monkey_i
       goober_j = monkey_j
       goober_k = monkey_k
    end procedure

The local variable goober is ok, but the variable monkey will cause
problems.  I would have to up the intelligence of the processor to handle
those kinds of variables.

> Anyways... nice work.

Thanks

> Ralf



 _______  ______  _______  ______
[    _  \[    _ ][ _   _ ][    _ ]
[/| [_] |[/| [_\][/ | | \][/| [_\]
  |  ___/  |  _]    | |     |  _]
[\| [/]  [\| [_/] [\| |/] [\| [_/]
[_____]  [______] [_____] [______]
xseal at harborside.com  ICQ:13466657
http://www.harborside.com/home/x/xseal/euphoria/