improvement of literal run length & match length encoding about lzsa

[specke]

In fact, there is another way to improve the format, which will be almost compatible with the existing format. The pseudo-code would look as follows:
if (len == low) {
len += byte1;
if (len <= 255)
// same as before, +1 byte for lengths low..255
else if (len == 256)
// same as before, +3 bytes for lengths up to 65535
else
// we used +2 bytes for lengths 256..511
// however, we can use (len-256)*256 + byte2 which will make use of all currently unused values.
}

In fact Z80 decompressors are already compatible with this format change. However, current decompressors for 8088 and 6502 are not compatible. Our current plan is to modify decompressors for this change and introduce the change at the compressor level together with the next minor version update. Or the one after. We do not want to rush it, esp. since the change in the compression ratio is basically negligible.

[specke]

Hi da0ka, yes, this is a small inefficiency in the format, which was introduced to save on an extra operation in the decompressor. I.e. it is a performance optimization. Since the compressor itself is nearly optimal, it very well compensates for this inefficiency, literally losing 2 or 3 bytes per file usually.