Arithmetic coding in Fedora libjpeg (bug #639531)

[Kevin Kofler]

Björn Persson wrote:
> It's well known around the Internet that to achieve compatibility you
> should be conservative in what you send and liberal in what you accept.
> Applied to JPEG: Use only Huffman coding when encoding – except maybe if
> you know that all recipients can handle arithmetic coding – but support
> both encodings when decoding.

+1

I can see not adding the support to the ENCODING portion, also because that 
appears to imply an ABI change (extra flag), but the DECODING portion of the 
patch should be merged.

The patch against the original libjpeg libjpeg-turbo is derived from should 
probably work, it might not be "turbo"-optimized, but given how rare those 
files are, that shouldn't be a problem. It's still better to be able to 
decode the files slowly than not at all.

        Kevin Kofler