Not an answer to your question, but instead of using the Cryptix provider I'd go with http://www.bouncycastle.org/ or the AES implementation of JCE itself. (Rijndael became AES, but I'm sure you know that already.)
This page suggests that the JCE implementation can use AES combined with ECB. [ March 11, 2008: Message edited by: Ulf Dittmer ]
At my current skill level, I picked the Cryptix provider because I liked the name better. I unpacked all the sources because I can copy-paste faster from them than from the api-docs, which run somewhat lumpy to me because of running in the browser. It is my personal opinion that the naming of the providers reflects political leanings and I would like to stay with Crypt Ticks. Their docs state NoPadding and None, the page you cite uses PKCS7Padding ECB does not need padding, does it ? At least not if it runs one byte at a time, which was my understanding from reading the provided examples.
The last html page I read from Sun plainly claims AES and so on are now avaiable without loading third party libs. I wanted to use ECB/NoPadding because it fits my understanding at the moment, I am trying to get into all the Prime Number Generation / Vol 4 Knuth and everything but that is slow trudgery for me now. Use of the AES implementation of JCE itself. suits my needs well, I just used the Cryptix because when I got to being able to code my first DES-56 ( I mean really first ) that was where I was able to read an example and copy-paste directly from source code.
I kinda like the Cryptix comments better also. I changed the name from AES to "Rijndael" as a consequence of reading this week that AES is Rijndael - not because I knew what difference it would make. I had sample code that accepted the algorithm name Rijndael, could not find "AES" the provided sources. Let me clarifiy I should be doing 32-bit beginner shift/XOR right now ~ I am going to 2048/AES because the people who will be reviewing the code will not take me seriously with any less.