Re: posix atime/ctime changes despite mtime being set

[Piotr Łobacz]

Ooooo jesus i have discovered it finally.

I have verified two different directories which are suppoused to be packed, for this packages in reproducibleA and reproducibleB and to my supprise the difference lies in there from the beginning, even before the creation of this packages with tar command in opkg-build script...

Now, I need to investigate it.

BR

Piotr

Wysyłane z aplikacji Outlook dla systemu iOS

---

Od: bug-tar-bounces+p.lobacz=welotec.com@gnu.org <bug-tar-bounces+p.lobacz=welotec.com@gnu.org> w imieniu użytkownika Piotr Łobacz <p.lobacz@welotec.com>
Wysłane: Monday, July 31, 2023 8:28:22 PM
Do: Sergey Poznyakoff <gray@gnu.org.ua>; Paul Eggert <eggert@cs.ucla.edu>
DW: bug-tar@gnu.org <bug-tar@gnu.org>
Temat: Re: posix atime/ctime changes despite mtime being set

 

OK, so for the acpid-src package builds has been done and to my suprise the $build_date variable is equal 1663291482 and it is exactly the same for both builds.

I am more and more inclined towards Sergey's suggestion that the gormat changes in some magical way, but how? Or maybe some additional parameter is being added which we don't see?

Regarda,

Piotr

Wysyłane z aplikacji Outlook dla systemu iOS

---

Od: bug-tar-bounces+p.lobacz=welotec.com@gnu.org <bug-tar-bounces+p.lobacz=welotec.com@gnu.org> w imieniu użytkownika Piotr Łobacz <p.lobacz@welotec.com>
Wysłane: Monday, July 31, 2023 7:47:25 PM
Do: Sergey Poznyakoff <gray@gnu.org.ua>; Paul Eggert <eggert@cs.ucla.edu>
DW: bug-tar@gnu.org <bug-tar@gnu.org>
Temat: Re: posix atime/ctime changes despite mtime being set

 

 > Piotr, does that seeseem plausible?

No, because it uses hosttools which are on my local machine, but I can add in further tests a print to file with tarformat variable.

P.S. Sorry for my long responses but currently the build is running and it takes 10h to reproduce it and during that time my pc is not usable at all, that is why I'm responding from my mobile...

BR

Piotr