Working on an overlay compressed filesystem (using fuse).
Now I have the "non compression" part down and working, but... running into some problems.
I need a compression library that'll allow me to:
A : know the size of the original file (so I can return that when software asks for filesize, if not read's and write's to file will screw up).
B : allow me to read specific parts of a file, not decode the entire thing every time.
Those two things would allow me to override read and write's with ease and make the codesize much more managable.
Got a feeling I wont find it, and I'll end up having to code something to cache the original filesize :/, if so I'll probably end up with zlib.
Really wish zlib's compressBound was accurate :/
Compression algorithms
Message
A can probably most easily be done in your own higher-level library. I doubt any compression libraries track this themselves. As for B, you'll need to read in since the last reset of the compression state in order to rebuild the compression state at the point you're reading from or writing to.A : know the size of the original file (so I can return that when software asks for filesize, if not read's and write's to file will screw up).
B : allow me to read specific parts of a file, not decode the entire thing every time.
For efficient random access you'll probably need to break a file into a number of fixed size, independently compressed blocks, and delay some time before actually compressing them to combine writes. Not to mention find a way to pack these while minimizing slack space.