Here is a simple table for MySQL:

CREATE TABLE `aaaa` (
 `test` double unsigned NOT NULL
) ENGINE=MyISAM DEFAULT CHARSET=utf8

But when I convert it into a db_schema.dba, I get:

<?php
// This array contains the database schema
$schema = array (
  'aaaa' => 
  array (
    'name' => 'aaaa',
    'fields' => 
    array (
      'test' => 
      array (
        'type' => 'double unsigned',
        'not_null' => '1',
        'default' => false,
      ),
    ),
    'indexes' => 
    array (
    ),
  ),
  '_info' => 
  array (
    'format' => 'generic',
  ),
);

?>

=> The type is parsed as 'double unsigned', it should be 'double' only

It comes from the regex into eZMysqlSchema::parseType, which is currently:

preg_match ( "@([a-z ]*)(\(([0-9]*|[0-9]*,[0-9]*)\))?@", $type_info, $matches );

with the [a-z ], it supposes a type can contain a space! I can assure you it's impossible (you can check all types listed into phpMyAdmin or look -r into http://dev.mysql.com/doc/refman/5.0/en/data-types.html)

Never a mysql type will contain a space, so you should delete it.
Further more, I remember (but not sure) in the past, mysqldump was using uppercases type, and in fact phpMyAdmin shows them as uppercase

=> so, for a security, I would let a-z and A-Z for parsing (hence @i modifier)

code should be:

preg_match ( "@([a-z]*)(\(([0-9]*|[0-9]*,[0-9]*)\))?@i", $type_info, $matches );

a pull request in a few moment...