Better Compression?

Nov 15, 2007 at 6:05 PM
I tested your compression against other commercial and non-commercial products. It doesn't do as well as I would like. ;-)

Are you looking into improving this? Or is this as good as it gets.


Dec 6, 2007 at 7:24 PM
2 things:
  • the basic deflate compression algorithm is built into the .NET Framework base classes. It is as good as it gets, for now. This Zip library relies on the built-in class for the deflate, and augments that with a class that handles the zip-file format - header data, reading and writing the zip records, etc.

  • there is an optimization I made in the latest version that does not compress if the deflate causes an increase in the file size. This might happen when applying the deflate transform to a previously-compressed block, let's say a jpg file or another zip file. Previously when this happened the "deflated" (yet expanded) block was included in the zip. Now, the code compares the before and after compression sizes and includes the smaller of the two. So a file like a jpg won't be compressed, but also it won't expand.

So this is sort of as good as it gets, for now.