IPF Library Changes

2010-05-13

Backwards gap decoding would require practically implementing half of the core for reversed stream/bit/decoding/encoding operations... which is not exactly an economical solution.

Instead, we reversed the order of gap descriptors generated by the analyser - this would enforce correct IPF code generation in forward gap order.

The only downside of the solution is that the stream has to be “fast forwarded” to the point where the decoding should start to end at exactly the expected location, i.e. right before the start of the next block - but once that is done the process is identical to forward gap generation re-using existing functionality.