[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: bzip2 -- no longer non-us?

>>>>> "Mark" == Mark Eichin <eichin@cygnus.com> writes:
Mark> Indeed. On the egcs snapshot, we save 2M out of 9.5M; it takes 2x as
Mark> long as gzip -9, 4x as long as the default gzip (-6 according to the doc.)
Mark> I believe it is worth considering this for install kits and possibly
Mark> for .deb internals [consider that we could leave the *control* element
Mark> gzip'ed, while bzip'ing the data element, in order to provide clean
Mark> upgrade warnings -- ie. the old dpkg could still parse the new format
Mark> and find the dependency on a new version of dpkg, or something like
Mark> that.  This is *in addition* to updating the debian-binary tag, I
Mark> think...] 
Mark> bzip2:    263.85user 1.17system 5:14.89elapsed 84%CPU
Mark> bzip2 -9: 265.51user 1.36system 5:31.26elapsed 80%CPU
Mark> gzip:      82.28user 1.85system 1:51.52elapsed 75%CPU
Mark> gzip -9:  133.57user 1.62system 2:44.22elapsed 82%CPU
Mark>   -rw-r--r--   1 eichin   cygnus    7467838 Aug 27 19:53 egcs-970825.tar.bz
Mark>   -rw-r--r--   1 eichin   cygnus    7467838 Aug 27 20:15 egcs-970825.tar.bz9
Mark>   -rw-r--r--   1 eichin   cygnus    9637129 Aug 27 20:01 egcs-970825.tar.gz
Mark>   -rw-r--r--   1 eichin   cygnus    9566095 Aug 27 20:07 egcs-970825.tar.gz9

However bzip2 is still too slow at uncompressing to be used in .deb
files IMHO.  Here are my tests on the same file (Sun Ultra 200MHz) :

		size		Compr. time 	Decompr. (s)
gzip -9		9566095		 74		 6
bzip -9		7386799		218		78
bzip2 -9	7467838		150		44
bzip2 -1	8852622		115		26

bzip2 is several times (4->8) slower than gzip, so users of slow
machines may not be happy.  And with the -1 option, savings are not so


TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .

Reply to: