Question about Reed Solomon decoder for 802.16

Hi,
I have a question about decoding Reed Solomon code used in 802.16
standard. In the 802.16-2004 document, RS(255,239,8) is used, but the
output is usually punctured to produce shorter codeword, RS(40,36,2) for
example. The question is: because the parity bytes is reduce from 16 bytes
to 4 (first) bytes, how we design the decoder? Padding zero bytes to the
parity and data to form RS(255,239,8) decoder? Wrong, because the 4 parity
bytes can correct the total of 12 parity bytes (0 padding) and (xxx) data
bytes.Or we must design a RS(40,36,2) decoder, but the code generation
polynomial is changed(?). 
Pls give me advice. 
Thank you.    

Sorry because I have asked such @#$%^& question. Just think again, I
realize that the decoder would be RS(255,239) with 12 erasures
corresponding to 12 cleared parity bytes (for RS(40,36,2) code), and it
can still correct 2 other errors. 
:D

>Sorry because I have asked such @#$%^& question. Just think again, I
>realize that the decoder would be RS(255,239) with 12 erasures
>corresponding to 12 cleared parity bytes (for RS(40,36,2) code), and it
>can still correct 2 other errors. 
>:D
>
>
>
How did you manage to undo the shortening? I thought reordering the
sequence and using zero padding would be enough, but the output is not
working. Thanks in advance.
Victor