Originally posted by skeevy420
View Post
Files stored on file systems have to be seekable, thus transparent compression algoritms typically compress files in very small blocks (because you cannot seek within a compressed stream). Between blocks, all compressor context is lost. Thus, if two variants of the same code are too "far" from each other, they will not benefit from dictionary compression (which is what should help the most with fat binaries, since most of the code will likely remain the same).
Comment