Re: the wonder of "cannot assign to arbitrary elements"

[Manuel Collado]

El 12/12/2021 a las 19:52, Denis Shirokov escribió:
> BEGIN{
>         typeof( a )
>         SYMTAB[ "a" ] = 1
>         SYMTAB[ "b" ] = 2                                     # <- here will be 
> an arbitrary element erro
> }

This is on purpose.

> could someone please inform me what's the difference between a and b ?
> and how to avoid an error with 'a' ?
------------------------------->-'b', I suppose.

Variables are created when the awk script is parsed, not when it is 
executed. "a" is the name of a variable, and hence there is a 
corresponding SYMTAB entry.

"b" is just an invalid index of SYMTAB. Invalid because there is no 
variable with such name.

> feature of accessing the content of the global variables for writes is useless.

You can *assign* values to variables via SYMTAB, but you can not use 
SYMTAB to *create* variables. To create a variable, just mention it in 
the script code.

gawk '
BEGIN{
  typeof( a )
  b
  SYMTAB[ "a" ] = 1
  SYMTAB[ "b" ] = 2    # <- No error here.
  print b
}'

2

Works as expected.

> commonly - the fatal "cannot assign to arbitrary elements" - is idea that's
> does more harm than good. it's practically nullifies feature of write
> access to the content
> of global variables because it is better to perform direct value
> writing instead of trying
> to understand why(and where) gawk determinates an element as an
> arbitrary element of SYMTAB.
>
> as a gawk user i may say that without fatal "cannot assign to
> arbitrary elements" - life was been really easier and nice
> and gawk was been more flexible.

It was in the past. But with a different, confusing semantics. So 
arbitrary SYMTAB indexes were finally forbidden.

> Kind Regrads
> Denis

HTH. Regards.

--
Manuel Collado - http://mcollado.z15.es