Consider the following:
>> x % x is a simple two-field structure
x =
x: 1 y: 'y'
>> type('printMe.m') % printMe is just a simple function
function printMe(x) size(x), x
>> printMe(x)
ans =
1 1
x =
x: 1 y: 'y'
>> printMe(struct('x', 1, 'y', 'y')) % Same result passing in an equivalent anonymous struct
ans =
1 1
x =
x: 1 y: 'y'
>> % But look what happens if one of the values is an empty cell array
>> printMe(struct('x', 1, 'y', {}))
ans =
0 0
x =
0x0 struct array with fields: x y
>> % Is this because it's empty
>> printMe(struct('x', 1, 'y', ''))
ans =
1 1
x =
x: 1 y: ''
>> printMe(struct('x', 1, 'y', []))
ans =
1 1
x =
x: 1 y: []
>> printMe(struct('x', 1, 'y', struct))
ans =
1 1
x =
x: 1 y: [1x1 struct]
>> % Apparently not!
So, apparently, if (and only if) a value in an anonymous struct passed to a function is an empty cell array, the function "loses" all the values in the anonymous struct and thinks it's empty (though, curiously, it doesn't completely "lose" the struct and in particular it retains "knowledge" of its fields, just not the values assigned to those fields).
Bug or "feature," and if the latter, rationale?
Thanks for your time,
OlyDLG
Best Answer