Hello,
I, for one, am no longer excited by bzip2 nowadays. Just made some
tests on a random mailing list archive:
Time(s) Size(B) Time vs. gz-9 Size vs. gz-9
Bzip2-6 25,17 31774760,00 2,8700114025 0,8093237242
Bzip2-9 29,90 29919222,00 3,409350057 0,7620619692
Xz-3 25,01 22065920,00 2,8517673888 0,562033279
Xz-4 45,39 20840460,00 5,1755986317 0,5308200188
Xz-5 61,77 19004876,00 7,0433295325 0,4840665051
Xz-6 96,38 18240560,00 10,9897377423 0,4645988814
Xz-9 116,74 15741240,00 13,3112884835 0,4009395816
Gzip-9 8,77 39260878,00 1 1
So it seems that xz is always better than bzip2 and considering the
computational costs of archivemail, xz at level 4 looks like a good
trade-off.
But it looks like we have to wait for python 3.3 to get native xz
support, see http://bugs.python.org/issue6715 for details.
Regards,
Eduard.
--
<Myon> 36m² ist ganz schön viel
<rvb> Hm, das sind mehr als 9m x 9m.
<rvb> Halt, nein ... 9x9 ist nicht 18.
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
