Re: [Qemu-devel] [PATCH] build: change dist target to use xz

[Brad Smith]

On Fri, Jan 04, 2013 at 08:03:05PM -0500, Brad Smith wrote:
> On Fri, Jan 04, 2013 at 05:41:41PM -0700, Eric Blake wrote:
> > On 01/04/2013 05:28 PM, Brad Smith wrote:
> > > On Thu, Aug 02, 2012 at 06:08:15AM -0400, Brad Smith wrote:
> > >> If a compression format other than gzip is used we might
> > >> as well move to xz instead of bzip2.
> > 
> > Agreed, many projects are dropping bzip2 as irrelevant (slower than the
> > more-widely-available gz, compresses worse than newcomer xz).
> > 
> > > +++ b/scripts/make-release
> > > @@ -20,5 +20,5 @@ git checkout "v${version}"
> > >  git submodule update --init
> > >  rm -rf .git roms/*/.git
> > >  popd
> > > -tar cfj ${destination}.tar.bz2 ${destination}
> > > +XZ_OPT=-9 tar cfJ ${destination}.tar.xz ${destination}
> > 
> > Do we really need -9?

ping.

> No. There isn't much of a difference between 8e and 9 with QEMU's
> tar archive..
> 
> 8175644 qemu-1.3.0.tar.xz_8e
> 8164188 qemu-1.3.0.tar.xz_9
> 
> > For example, coreutils did an interesting
> > benchmark in its cfg.mk file, and chose to use -8e instead of -9 as a
> > result:
> > 
> > >> # Comparing tarball sizes compressed using different xz presets, we see 
> > >> that
> > >> # an -8e-compressed tarball is only 9KiB larger than the -9e-compressed 
> > >> one.
> > >> # Using -8e is preferred, since that lets the decompression process use 
> > >> half
> > >> # the memory (32MiB rather than 64MiB).
> > >> # $ for i in {7,8,9}{e,}; do \
> > >> #     (n=$(xz -$i < coreutils-8.15*.tar|wc -c);echo $n $i) & done |sort 
> > >> -nr
> > >> # 5129388 7
> > >> # 5036524 7e
> > >> # 5017476 8
> > >> # 5010604 9
> > >> # 4923016 8e
> > >> # 4914152 9e
> > 
> > Also, your use of tar cfJ assumes relatively new GNU tar; it won't work
> > on other systems (but this is a pre-existing non-portability, and may be
> > irrelevant, if you are willing to require that the only people likely to
> > make a release tarball are using appropriate tools).
> 
> It is already using a GNU tar-ism with the j vs J flags and IMO GNU tar
> 1.22 at almost 4 years old has been around more than long enough.
> 
> 
> diff --git a/Makefile b/Makefile
> index a7ac04b..9073a86 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -251,10 +251,10 @@ clean:
>  
>  VERSION ?= $(shell cat VERSION)
>  
> -dist: qemu-$(VERSION).tar.bz2
> +dist: qemu-$(VERSION).tar.xz
>  
> -qemu-%.tar.bz2:
> -     $(SRC_PATH)/scripts/make-release "$(SRC_PATH)" "$(patsubst 
> qemu-%.tar.bz2,%,$@)"
> +qemu-%.tar.xz:
> +     $(SRC_PATH)/scripts/make-release "$(SRC_PATH)" "$(patsubst 
> qemu-%.tar.xz,%,$@)"
>  
>  distclean: clean
>       rm -f config-host.mak config-host.h* config-host.ld $(DOCS) 
> qemu-options.texi qemu-img-cmds.texi qemu-monitor.texi
> diff --git a/scripts/make-release b/scripts/make-release
> index 196c755..9b70650 100755
> --- a/scripts/make-release
> +++ b/scripts/make-release
> @@ -20,5 +20,5 @@ git checkout "v${version}"
>  git submodule update --init
>  rm -rf .git roms/*/.git
>  popd
> -tar cfj ${destination}.tar.bz2 ${destination}
> +XZ_OPT=-8e tar cfJ ${destination}.tar.xz ${destination}
>  rm -rf ${destination}

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.