If pixz is exited mid-compression, the whole archive is useless. about pixz

[philip-wernersbach]

I'd also like to add that I am willing to code this particular feature, but I don't understand pixz's file format enough to point pixz to the second to last file in this case.

[vasi]

Pixz uses the exact same file format as xz: http://tukaani.org/xz/xz-file-format.txt . The only difference is that xz usually creates single-block streams in its output files, while pixz creates multi-block streams.

Unfortunately the xz format stores a lot of important data at the end ("footer") of the file, so the file on-disk will never be a 100% valid xz file until the compression has finished. Also, pixz is a stream-based compressor—it doesn't compress one file at a time, it just compressed blocks of data. So if you do have a truncated xz file from pixz, it won't end after a file, but somewhere in the middle of a file.

That said, there should be enough information in the xz files produced by pixz to at least extract the data inside. Have you tried forcing pixz into streaming decompression mode? cat myfile.tar.pxz | pixz -d > myfile.tar That might produce something useful. You could also try using plain old xz to do your decompression, it might be more robust.

Or do you want to do more than just extract the data? If you want to continue from an interrupted compression, or to repair the result of an interrupted compression, that will be harder. I guess you could probably code up some sort of checkpointing feature, at the cost of slightly decreased performance.