OK, ignore my last msg. I have forgotten to pass --format=posix for extraction...
BR Piotr ________________________________ Od: bug-tar-bounces+p.lobacz=welotec....@gnu.org <bug-tar-bounces+p.lobacz=welotec....@gnu.org> w imieniu użytkownika Piotr Łobacz <p.lob...@welotec.com> Wysłane: czwartek, 3 sierpnia 2023 09:01 Do: Sergey Poznyakoff <g...@gnu.org.ua> DW: Paul Eggert <egg...@cs.ucla.edu>; bug-tar@gnu.org <bug-tar@gnu.org> Temat: ODP: posix atime/ctime changes despite mtime being set Hi Sergey, you are right but additionally this is not a tar archive. I have tested this command with additional params e.g. --format=posix: tar -I "pzstd -8 -p 16" -cS --format=posix --acls --xattrs --numeric-owner -f /home/plobacz/workspace/mpa_yocto/build/tmp-glibc/work/core2-64-WelotecGmbH-linux/acpid/2.0.33-r0/acpid-2.0.33-r0.tar.zst * and file says that it is: acpid-2.0.33-r0.tar.zst: Zstandard compressed data (v0.8+), Dictionary ID: None So, the whole command is broken. And even with posix format the timestamps are cut. BR Piotr ________________________________ Od: Sergey Poznyakoff <g...@gnu.org.ua> Wysłane: środa, 2 sierpnia 2023 07:14 Do: Piotr Łobacz <p.lob...@welotec.com> DW: Paul Eggert <egg...@cs.ucla.edu>; bug-tar@gnu.org <bug-tar@gnu.org> Temat: Re: posix atime/ctime changes despite mtime being set Hi Piotr, > do these tar commands in here > https://eur04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgit.yoctoproject.org%2Fpoky%2Ftree%2Fmeta%2Fclasses-global%2Fsstate.bbclass%23n859&data=05%7C01%7Cp.lobacz%40welotec.com%7Cde0d37ead42042f375b108db93175e09%7C25111a7f1d5a4c51a4ca7f8e44011b39%7C0%7C0%7C638265500806798082%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=CWFkI0RD2USu2CrQPbAMDGPjef9gZtdU3S8raC2NUSw%3D&reserved=0<https://git.yoctoproject.org/poky/tree/meta/classes-global/sstate.bbclass#n859> > may somehow be removing timestamps? Output archive format is not specified, therefore the default (gnu) is used. This format keeps timestamps with second precision. That's what we have suspected from the very beginning. If you intend to create reproducible archives, the right course of action would be to invoke tar with the same set of options. Regards, Sergey