Re: How to code a complex AoH?

[Athanasius]

Hello iatros, and welcome to the Monastery!

Each element of a hash (or of an array, for that matter) must be a scalar value. So if you want to store an array of values in the points slot of a hash, you have to store a pointer to that array:

$student = {
    ...
    points => \@points,
};

or

$student = {
    ...
    points => [@points],
};

See perlreftut and perldsc.

Update: To elaborate on stevieb’s point: if the array @points contains the elements ('a', 'b', 'c', 'd'), then the assignment

$student = {
    ...
    points => @points,
};

is effectively this:

$student = {
    ...
    'points', 'a',
    'b', 'c',
    'd',
};

or, equivalently,

$student = {
    ...
    points => 'a',
    b      => 'c',
   'd',
};

— which explains why the compiler is warning about an odd number of elements in the hash: the last array value ('d') becomes a hash key with no associated value.

Hope that helps,

[haukex]

points => \@points,

Actually, that won't work here - it looks like iatros isn't using strict, or has predeclared @points outside of the loop, so that \@points and therefore $$student{points} will always point to the same array, and that array will get overwritten on each iteration of the loop. points => [@points], will work correctly, since it creates a (shallow) copy of the array. (Athanasius, I know you know all this, the explanation is for the benefit of the OP.)

iatros: You really should Use strict and warnings, and then use my to declare your variables, including inside the loop: my ( $id , $gender , $birthday ... ) = .... Then you can use both of the code examples that Athanasius showed, because then on each iteration of the loop, @points will be a "new" array.

Update: Tweaked explanation a tiny bit.

[AnomalousMonk]

... the last array value ... becomes a hash key with no associated value.

iatros: Just another | a minor tweak to Athanasius's otherwise excellent ++explanation: The unpaired key will not have no associated value, but rather the undef value (that's where the  '.5' => undef in the OPed dump comes from). This, of course, will probably just lead to more warnings down the line!

[iatros]

Thank you for the reply. I hope I've learned my lesson with your help. I will think over the parsing thing. As Larry put it: TIMTOWTDI -HM-