Re: [bug-gawk] fatal: split: second argument is not an array

[Aharon Robbins]

Hi. Thanks for the note.

> Date: Thu, 21 Mar 2013 22:57:49 -0300
> From: Hermann Peifer <address@hidden>
> To: address@hidden
> Subject: [bug-gawk] fatal: split: second argument is not an array
>
> Hi,
>
> I tried to split fields 1, 2 and 3 into arrays a[1], a[2] and a[3] 
> respectively, see the below example (1). I ended up with the 
> above-mentioned fatal error and I wonder if the error is correct as the 
> identifiers a[1], a[2] and a[3] have never been used before. So in my 
> (potentially wrong) understanding, they are "untyped", which means they 
> could be either used as a scalar variable or as an array.

I'm afraid you're misunderstanding. Any reference to a previously 
non-existent element creates it *as a scalar* with an untyped value (which
acts as the null string or zero).

Consider:

        BEGIN { printf("a[1] is <%s>\n", a[1]) }

a[1] is created and given an empty value, but it's a scalar value. This
is how awk has worked since The Beginning....

> I got around the fatal error by initialising a[1], a[2] and a[3] as 
> arrays, see (2), but I wonder if this hack should really be necessary.

Yes, it is necessary.

Thanks,

Arnold