On 11/14/2017 12:37, L A Walsh wrote: > Tetsuji Rai via cygwin wrote: >> Hi all, >> >> I have many music files, so I often archive music directory with "tar >> -cJvf music.tar.xz music" with environment variable XZ_OPTS="-T 3" (I >> have 4 threaded first generation i7). The archive is about 30GB. >> Until some weeks ago it worked fine. However since I updated to the >> latest version of cygwin64 some weeks ago, I have a trouble extracting >> from that archive at some point. "tar -xJvf music.tar.xz" says "archive >> file is corrupt" or something and quits. T > --- > Have you tried it with one xz thread?... > > BTW to compress/decompress w/xz, I usually use > tar caf /tmp/archive.txz /music_archive to compress and > tar xaf /tmp/archive.txz to uncompress > > (but those should be unrelated to your problem). > > I just tried "-T 3" on my version of tar and with "xz" > and it doesn't make a difference in time, but does result > in slightly larger file sizes -- which is characteristic > of it using multiple threads. However, tar only produces > 1 thread, so it never gets a chance to run things in multiple > threads. > > Also -- > > Besides, music isn't "real compressible" -- especially > if you are compressing mp3 or flac, but the general purpose > compression algorithms don't do a great job of compressing > music. > > Compressing mp3->mp3+xz saved about 2.8% > Compressing flac->flac+xz saved about 0.82% (<less than 1%) > > Compressing raw waves: > xz files were 32% LARGER than flac. > > So what type of music are you compressing? > > BTW -- sending this to you in addition to list, since > the list has been flakey for me recently -- seems to > have a bug rejecting my domain. > > > Now I found single compression doesn't harm at all. Multithreaded compression is the problem. Actually, compression of mp3 files doesn't mean anything.....so I will just use tar for archive. It's much faster :)
-- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
