lh2 and lh3

[polluks]

• lh2

  lh1 variant. This method supports 8 KiB sliding window, with support of maximum 256 bytes of matching length. Dynamic Huffman encoding is used.

• lh3

  lh2 variant with Static Huffman.

https://en.wikipedia.org/wiki/LHA_(file_format)#Canonical_LZH

[fragglet]

If any examples of these archives exist in the wild then I'd love to support them. However, I'm pretty sure there aren't any. I don't really want to support these formats without test archives that can be used for regression testing. If you find any tools that can generate lh2 or lh3, let me know.

[roytam1]

these files can be created with unlha32.dll (-jm5 for -lh2-, with file size limit 8KB, and -jm6 for -lh3-) you can use undll.exe (http://k-takata.o.oo7.jp/mysoft/undllexe.html) and rename undll.exe as unlha32.exe and put unlha32.dll (from http://www.madobe.net/archiver/lib/unlha32.html) aside to use.

and created lzh files are test OK with LHA32 2.67.00.

command to create lh2.lzh:

unlha32 a -jm5 lh2.lzh LICENSE.MIT

command to create lh3.lzh:

unlha32 a -jm6 lh3.lzh LICENSE.MIT

lh2-lh3.zip source of LICENSE.MIT: https://raw.githubusercontent.com/theupdateframework/specification/master/LICENSE-MIT.txt

[roytam1]

I think I should tag #11 as well

[roytam1]

-jm5 for -lh2-, with file size limit 8KB

this limit respects original LH(arc) 1.9x original implementation, but can be bypassed by hex-editing unlha32.dll (I used unlha32.dll v3.00 as example) search "72 03 6a 03 5f 33" and change "72" to "eb" to bypass original file size check.

[fragglet]

Any idea why there's an 8KB file size limit?

[roytam1]

Any idea why there's an 8KB file size limit?

maybe because of processing power in 1989 is not that capable for generating Dynamic Huffman encoding for larger than 8KB.