On Dec 19, 3:29pm, mar...@duskware.de (Martin Husemann) wrote:
-- Subject: Re: update on reproducible builds
| On Mon, Dec 19, 2016 at 09:24:00AM -0500, Christos Zoulas wrote:
| > Because we want keep the checksums of all the files identical. Since some
| > files (like tar files or
On Mon, Dec 19, 2016 at 09:24:00AM -0500, Christos Zoulas wrote:
> Because we want keep the checksums of all the files identical. Since some
> files (like tar files or random archives contain timestamps, this is
> impossible
> without choosing a particular timestamp).
Yes, I understand, but I dis
On Dec 19, 12:28pm, mar...@duskware.de (Martin Husemann) wrote:
-- Subject: Re: update on reproducible builds
| On Mon, Dec 19, 2016 at 12:19:51PM +0100, Thomas Klausner wrote:
| > This was already possible before -- you set MKREPRO=yes and
| > MKREPRO_TIMESTAMP to the seconds-since-epoch o
On Mon, Dec 19, 2016 at 01:47:54PM +0100, Thomas Klausner wrote:
> Because the files are tarred up into release tarballs, and ideally,
> those tarballs would be bit-identical if built from the same sources.
Yes, but I'd consider that fully cosmetic.
Martin
On Mon, Dec 19, 2016 at 12:28:55PM +0100, Martin Husemann wrote:
> On Mon, Dec 19, 2016 at 12:19:51PM +0100, Thomas Klausner wrote:
> > This was already possible before -- you set MKREPRO=yes and
> > MKREPRO_TIMESTAMP to the seconds-since-epoch of that timestamp.
>
> I fully fail to understand
>
On Mon, Dec 19, 2016 at 12:19:51PM +0100, Thomas Klausner wrote:
> This was already possible before -- you set MKREPRO=yes and
> MKREPRO_TIMESTAMP to the seconds-since-epoch of that timestamp.
I fully fail to understand
(a) the desire to keep the timestamps all identical (but that is
more a
On Mon, Dec 19, 2016 at 06:17:13PM +0700, Robert Elz wrote:
> When I do real builds (as opposed to "does it still build after
> I just mangled things" builds) I do an update with -D (or -r:
> if it isn't a -current build but a build from a branch) using a specific
> timestamp (most commonly the mo
Date:Sun, 18 Dec 2016 23:17:41 -0500
From:chris...@zoulas.com (Christos Zoulas)
Message-ID: <20161219041741.d824f17f...@rebar.astron.com>
| This timestamp is determined in the build system as the timestamp
| of the latest file committed in the source tree.
When I
In article <20161219041741.d824f17f...@rebar.astron.com>,
Christos Zoulas wrote:
>
>I have changed the CVS server on cvs.netbsd.org, not to do this
>anymore and always provide the timestamps to the client, which
>means that even "cvs update" now sets the time for updated files.
>I am planning to p
Hello,
Since yesterday the releng build server has been updated to run
"build.sh -P", which sets MKREPRO and MKREPRO_TIMESTAMP automatically.
For those not familiar with those options, the first variable
arranges things so programs don't contain build-specific dates/times
etc.; the second variabl
10 matches
Mail list logo