Re: --set-mtime-command (was Re: posix atime/ctime changes despite mtime being set)

[Paul Eggert]

On 2023-08-02 11:09, Sergey Poznyakoff wrote:

> > Some time ago Bruno Haible suggested a simpler approach, which may be
> > better for GNU tar's users. Here's the idea, which requires two passes
> > over the input files, the first to collect timestamp info:
>
> I like the idea, although I doubt that its implementation would be less
> complicated than the above.

Yes, it'd be more work for Tar. Less work for Tar's users, I hope.


> So we assume that --reproducible-mtime option would work only for git,
> right?

It should work for any version-control system. It doesn't even assume a 
version control system, or that you use a timestamp-based build system 
like 'make'. All it assumes is that you have source files maintained by 
hand with reliable timestamps, and that you did a build sequentially 
from scratch with an increasing clock so that build products' timestamps 
are in ascending order and have timestamps greater than the newest 
source file.


> In short, it looks like we have to find a better way to extract mtimes
> from git.

One possibility is to parse the output of the command "git log --raw 
--pretty=format:%ct". Admittedly this will be a bit of work.