Re: equivalent for C-style init: structname varname[] = {...} ?

[Sergei Steshenko]

----- Original Message -----
> From: Dimitri Maziuk <address@hidden>
> To: Sergei Steshenko <address@hidden>
> Cc: "address@hidden" <address@hidden>
> Sent: Monday, November 19, 2012 5:34 PM
> Subject: Re: equivalent for C-style init: structname varname[] = {...} ?
> 
> On 11/18/2012 11:42 AM, Sergei Steshenko wrote:
>> 
>>  "have mixed-type fields" - "foo" and "foo 
> bar" are _both_ strings, so it's the same type.
>> 
>>  I suggest to read http://en.wikipedia.org/wiki/Hash_table first.
>> 
>>  "two" and 2 would be different types (string vs (integer) 
> number), but it's not what I was talking about.
> 
> But that was what *I* was talking about:
> {
> string name;
> float value;
> }
> which an array or map does not allow. So octave structs are about as similar 
> to 
> structs as Pink Floyd is similar to Cherry Garcia ice cream.
> 
> Any yes, if they're hashmaps then obviously "foo bar" is a valid 
> map key, but "a( 3 ).foo bar" is not a valid syntax. Because the space 
> between foo and bar is a token separator.
> 
> Dima
>

I never meant to use non-quoted non-bareword key as field name.

However, both

1)

"
octave:2> foo = struct
foo =

  scalar structure containing the fields:


octave:3> foo."foo bar" = 1
parse error:

  syntax error

>>> foo."foo bar" = 1
                ^

octave:3>   
",

2)
"
octave:3> foo = struct
foo =

  scalar structure containing the fields:


octave:4> foo.("foo bar") = 1
error: invalid structure field name `foo bar'
octave:4>   
"

should work exactly because "foo bar" is a valid hash key.


And, as shown before, the key is accepted through 'setfield':

"
octave:4> foo = struct
foo =

  scalar structure containing the fields:


octave:5> setfield(foo, "foo bar", 1)
ans =

  scalar structure containing the fields:

    foo bar =  1

octave:6>     
".



Regards,
  Sergei.