Re: a bug or a feature?

[Aharon Robbins]

Greetings.  Re this:

> Date: Fri, 25 Apr 2003 11:59:25 +0200 (CEST)
> From: Boguslaw Jackowski <address@hidden>
> To: address@hidden
> cc: Martin Berggren <address@hidden>,
>    Janusz Marian Nowacki <address@hidden>
> Subject: a bug or a feature?
>
> Dear colleagues,
>
> We've encountered a strange behaviour of gawk which has something
> to do with the `delete' operator. The following three-liner
>
>   BEGIN {a(); b(S)}
>   function a() {delete S}
>   function b(s) {s[0]}
>
> reports an error:
>   ... fatal: delete: illegal use of variable `S' as array
>
> This error is reported under both Linux and Windows since the
> version (at least) 3.1.0 (also in 3.1.2).
>
> Apparently, gawk presumes that `S' should be a scalar judging from
> the statement `b(S)'. There is a simple circumvention---it suffices
> to add, e.g., a dummy function definition at the beginning of the script:
>
>   function c() {S[0]}
>
> Because of the weirdness of the circumvention, we doubt whether
> this behaviour can be considered a `feature'.
>
> Looking forward to hearing from you --
>
> yours sincerely, Bogus\l{}aw Jackowski, Piotr Strzelczyk
>
> Ps. Due to the optimization efforts in most recent versions
>     of gawk, the error is reported also for a `traditional'
>     form of using the `delete' operator:
>
>     function a() {for (i in S) delete S[i]}
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
>  Bogus\l{}aw Jackowski: address@hidden
> ----------------------------------------------------------------
>  Hofstadter's Law: It always takes longer than you expect, even
>                    when you take into account Hofstadter's Law.
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

This area is a bit messy in current gawk.  I have fixed things in
my current sources, which will eventually be 3.1.3.  For now,
you may use the patch Stepan posted, or use your workaround.

Thanks,

Arnold