> On Jan 11, 2016, at 3:35 PM, Richard Purdie > <richard.pur...@linuxfoundation.org> wrote: > > On Mon, 2016-01-11 at 15:21 -0800, Khem Raj wrote: >>> On Jan 11, 2016, at 2:36 PM, Paul Eggleton < >>> paul.eggle...@linux.intel.com> wrote: >>> >>> On Mon, 11 Jan 2016 13:17:14 Khem Raj wrote: >>>>> On Jan 11, 2016, at 1:07 PM, Paul Eggleton < >>>>> paul.eggle...@linux.intel.com> >>>>> wrote:> >>>>> On Mon, 11 Jan 2016 09:36:36 Paul Eggleton wrote: >>>>>> On Mon, 11 Jan 2016 09:26:39 Paul Eggleton wrote: >>>>>>> On Fri, 08 Jan 2016 18:22:49 Richard Purdie wrote: >>>>>>>> xz gives better compression results than bzip/gz but is >>>>>>>> often slower. >>>>>>>> Using parallel compression mitigates this somewhat and is >>>>>>>> particularly >>>>>>>> useful for the SDK. >>>>>>>> >>>>>>>> Signed-off-by: Richard Purdie < >>>>>>>> richard.pur...@linuxfoundation.org> >>>>>>>> >>>>>>>> diff --git a/meta/recipes-support/pixz/pixz_1.0.6.bb >>>>>>>> b/meta/recipes-support/pixz/pixz_1.0.6.bb new file mode >>>>>>>> 100644 >>>>>>>> index 0000000..e6e4ac2 >>>>>>>> --- /dev/null >>>>>>>> +++ b/meta/recipes-support/pixz/pixz_1.0.6.bb >>>>>>>> @@ -0,0 +1,14 @@ >>>>>>>> +SUMMARY = "Parallel, indexed xz compressor" >>>>>>>> + >>>>>>>> +DEPENDS = "xz libarchive" >>>>>>>> + >>>>>>>> +SRC_URI = >>>>>>>> "https://github.com/vasi/pixz/releases/download/v${PV}/${ >>>>>>>> BPN}-${PV}.tar >>>>>>>> . >>>>>>>> xz >>>>>>>> " >>>>>>> >>>>>>> Can we rely on this never changing? I thought we'd >>>>>>> experienced problems >>>>>>> with github's release tarballs being generated on the fly >>>>>>> in the past... >>>>>> >>>>>> Another thing, this seems to fail to build without asciidoc: >>>>>> >>>>>> ------------ snip ------------ >>>>>> checking for src/pixz.1... no >>>>>> checking for a2x... no >>>>>> configure: error: AsciiDoc not found, not able to generate >>>>>> the man page. >>>>>> ------------ snip ------------ >>>>>> >>>>>> This is also related to not supporting B != S, since >>>>>> src/pixz.1 does >>>>>> exist, >>>>>> just in S and not B. If you inherit autotools-brokensep >>>>>> instead of >>>>>> autotools it works. >>>>> >>>>> Possibly a bit obvious, but even inheriting autotools-brokensep >>>>> isn't >>>>> enough, because if it runs "make clean" on re-executing >>>>> do_configure, >>>>> src/pixz.1 gets deleted and you get the same issue. >>>> >>>> Adding --without-manpage might get you past this issue. >>> >>> I'm afraid that's not a valid option for this configure script. >>> >>> I had more shenanigans trying to build the target version. In the >>> end I needed >>> to add "ac_cv_file_src_pixz_1=yes" to EXTRA_OECONF and inherit >>> pkgconfig. >>> >> >> I think you are missing >> https://github.com/vasi/pixz/commit/936d8068ae19d95260d3058f41dd6cf71 >> 8101cd6 >> >> which is committed after 1.0.6 release. It should be back ported. >> may be not use tarball but straight use >> SRCREV=“936d8068ae19d95260d3058f41dd6cf718101cd6” >> with git fetcher might be better. > > As Randy mentions, should we use the -T option to xz instead though?
that might be ok too xz /tmp/xx.tar 19.65s user 0.07s system 100% cpu 19.714 total XZ_DEFAULTS="-T 0" xz /tmp/xx.tar 22.35s user 0.49s system 360% cpu 6.340 total pixz /tmp/xx.tar /tmp/xx.tar.xz 26.56s user 0.45s system 456% cpu 5.917 total a little slower than pixz and doesnt seem to use all cores with pixz I had 450% CPU on same load. using pixz is simpler though. It can also be called integrated with tar. for XZ we need to set this env variable to enable parallelism with pixz its not needed. > > Cheers, > > Richard
signature.asc
Description: Message signed with OpenPGP using GPGMail
-- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
