Daniel Kahn Gillmor dkg at fifthhorseman.net
Wed Jul 22 17:17:49 CEST 2015

On Wed 2015-07-22 12:57:36 +0200, Werner Koch wrote:
> On Wed,  8 Jul 2015 19:17, dkg at fifthhorseman.net said:
>> If removal of BUILD_TIMESTAMP isn't acceptable for gpgme, we should
>> allow it to be externally set for people who are trying to rebuild the
>> package reproducibly.
> Last year you added this to libgpg-error:
>     This patch allows an external build system to set this embedded
>     timestamp explicitly, which appears to make the package build
>     repeatably when ./configure is called with (for example)
>     --enable-build-timestamp=2014-09-23T01:02+0000
> The default is to use the actual time for the build, using
> --disable-build-timestamp will use "<none>" for the timestamp.  We
> should use a default for all packages.
> At least for Windows the build timestamp is useful for debugging and
> thus I would like to keep it.  However, changing the default to <none>
> is also fine with me.  For Gpg4win I would then change the default.
> Shall I change all packages to this scheme?

I'd be happy to have --disable-build-timestamp as the default for all
packages.  Thanks for the consideration!


More information about the Gnupg-devel mailing list