Ot: Special Text Compression For Old Computers?

Hi! First off, I have to apologize profusely for this way off-topic post when I should be asking on another forum. I am very close to being banned from two and was banned from another years ago for my bad behavior. I am working on a text compression program for 8-bit computers called PrintTok2. It was meant for text adventures on low-end computers, many having no more than 16k RAM. It is also designed such that a string can't rely on previous strings to aid in decompression. It currently uses tokenization and a modified version of Z-Machine's 5-bit-per-pixel approach. The modifications include an extra bit on some literals to add more character definitions and prevent the need to change dictionaries on them, an extra bit on dictionary-switches to determine whether they are for just the next character or several characters and the removal of an extra bit per word that determines if it's the last word. A version without literals compress is also on the way. I asked before, but the only help I got was a suggestion about Adaptive Huffman codes, but they would require access to previous strings during decompression. I believe I'm doing very well but want more ideas. Does anybody here have some?

 

BTW, I'm also working on general-purpose compression for both older and modern systems. One of my compression techniques uses AdaptHuff, LZ77, my Last16 technique which shortens repeated LZ77 blocks to a number indicating how many times ago it occurred, a sort of modified version of BPE which provides the offset of the previous repetition while skipping other repeats, ways of offsetting from latest LZ77 repeats, a modified version of Elias, an idea to not compress AdaptHuff values that are uncompressible and a sort of shortener if a byte was repeated one or two bytes ago. I also sometimes update the Huffman tree for blocks compressed by LZ77 and others if it works. Right now, my rendition of AdaptHuff is very buggy, and I need to fix it. I'm also aware of RLLZ, which was discussed on the comp.compression Usenet (?) group before they were closed. If you have any ideas, I'm all ears.