RE: [Firebird-devel] gpre coredump with CVS build

[Dimitry S. <SD...@to...>]

On 14 Oct 2003 at 7:13, Claudio Valderrama C. wrote:

>Maybe the long term solution is to make all callers aware that they can't
>request zero elements in the array, but they are logically correct: they
>request some node types that don't have arguments, hence they send zero.
>It's that #$%^&* design throughout the code that defines the first element
>inside the parent node that hurts us.

  That's why I suggested to use zero-length arrays, but the 
suggestion was rejected as non-standard.

   SY, Dimitry Sibiryakov.