gawk devel.- processing arrays of arrays

[Manuel Collado]

Reposted (sorry, already sent to the wrong address - please forgive me 
if you get duplicates)
---------------------

The forthcoming gawk 4.0 supports arrays of arrays. But it seems that 
there is no way to write generic code that process all (scalar) elements 
of a generalized array without knowing its structure in advance. The 
expected recursive construction could be:

function process_array( a ) {
   for (i in a) {
      if (isarray(a[i])) {
         process_array( a[i] )
      } else {
         ... process a[i] as scalar ...
      }
   }
}

Regrettably, there is no 'isarray()' primitive. I've thought about using 
'length(array)', but it is not array-specific (gives string length for 
scalars).

So please consider adding a new primitive to gawk-4.x. Perhaps either:
- isarray(x), or
- alength(x)
x been a scalar or an array - alength(scalar) should give a negative value.

The second alternative could render the length(array) extension unnecessary.

Am I missing something already available? Is there another way of 
enumerating all final elements of a multi-array without knowing its 
particular structure in advance?

Regards.

PS: Would it be appropriate to discuss this issue in comp.lang.awk?
--
Manuel Collado - http://lml.ls.fi.upm.es/~mcollado