This is not a critical problem, as in the dump the array key is saved for each index element. This is in turn used to recreate the index with elements in the proper order.

It is just a problem when you use DIFF to compare .dba files.

To explain it better: here's (part of) the output of ezsqldumpschema:

'indexes' =>
array (
'PRIMARY' =>
array (
'type' => 'primary',
'fields' =>
array (
1 => 'version',
0 => 'contentobject_attribute_id',
),
),
),

as you can see element 1 is first, element 2 is second