A kernel.org update

Posted Oct 31, 2013 13:05 UTC (Thu) by dlang (guest, #313)
In reply to: A kernel.org update by mgedmin
Parent article: A kernel.org update

> What's the story behind the phasing out of bz2?

If you are asking why:

IIRC, .bz2 is slower and larger than .xz (lzma), so it has no advantage

If you are asking about when, I can't answer

to post comments

A kernel.org update

Posted Oct 31, 2013 13:58 UTC (Thu) by Jonno (guest, #49613) [Link]

> IIRC, .bz2 is slower and larger than .xz (lzma), so it has no advantage
While there might be some pathological cases, in the general case:
xz -2 compresses faster, and usually better, than bzip2 -9.
xz -3 compresses better, and usually faster, than bzip2 -9.
xz always decompresses faster than bzip2.

The decompression memory requirement of xz -2 (3 MiB) and -3 (5 MiB) are comparable to bzip2 -9 (4 MiB), but the compression memory requirement are slightly higher (17M / 32M compared to 8M).

xz -6 (the default) compresses even better, but compress is usually about an order of magnitude slower than bzip2 -9, and it requires a lot more memory (94 MiB to compress, 9 MiB to decompress). So xz -6 is not really a direct replacement for bzip2, but can make sense when distributing files that are compressed once but downloaded and decompressed many times.

A kernel.org update

Posted Oct 31, 2013 14:32 UTC (Thu) by mgedmin (guest, #34497) [Link]

Thanks, that answers my question perfectly.