stupid me (Re: adaptive huffman, ordinary data)

>
> I think, hell, text should compress better than that, given a lot of chars
> are unused (many in the 0..31 range, 127..255), I would think I would be
> looking at at least 50%, oh well...
> it is possible though that the rle compressor adds some overhead, but it
> should be insignifigant here.
> possibly this could be related to me varying the tree construction a
> little...
>
me thinking it ammusing that no one caught me on this...
realizing that, just eliminating the values that wont occure would only give
marginal compression, thus what I am getting seems reasonable.

I then started writing a shannon-fano encoder to compare against my existing
ones, and roughly decided things were getting pointless (the encoder would
likely be slow given the manner of which the trees were being built and the
data encoded...).


as far as speed goes, my plain static encoder goes at a rate of about
3.72MB/s on my computer (athlon xp 1800+ pc133 sdram). I guess this is
decent. optimization is possible.

my dynamic encoder is are much slower, at about 1.65MB/s.

lz77 costs a lot as well, and beyond a certain amount I can't seem to boost
compression.


I am probably going to be off to different manner of tasks though.