On Mon, Oct 18, 2021 at 3:41 PM Bin Meng [off-list ref] wrote:

On Mon, Oct 18, 2021 at 6:08 PM Anup Patel [off-list ref] wrote:

quoted

On Mon, Oct 18, 2021 at 1:42 PM Bin Meng [off-list ref] wrote:

quoted

On Mon, Oct 18, 2021 at 3:44 PM Andreas Schwab [off-list ref] wrote:

quoted

On Okt 18 2021, Bin Meng wrote:

quoted

If this BUILD_DATE_EPOCH is not needed at all, I am not sure why do we
still keep

BUILD_DATE_EPOCH is also non-standard.  The standard variable is
SOURCE_DATE_EPOCH.

Yep, it's SOURCE_DATE_EPOCH in v4, not sure why Anup changed the name in v5.

Yes, I changed it to BUILD_DATE_EPOCH to imply that it is related to
BUILD_INFO. I was not aware that SOURCE_DATE_EPOCH is a
standard name. I will rename it to SOURCE_DATE_EPOCH in v6.

quoted

SOURCE_DATE_EPOCH is well documented and is the way for distros to
generate "reproducible builds".

I would vote to completely drop the "BUILD_INFO=y" as it is not necessary.

I don't agree with this. The BUILD_INFO helps us totally disable build
information whereas SOURCE_DATE_EPOCH helps us override the
build date/time when BUILD_INFO=y.

But what benefits us to provide the capability to conditionally
disable the build information? I think that is useful and can be
turned on unconditionally, and we can still generate reproducible
builds with SOURCE_DATE_EPOCH.

Users will have to figure-out/remember a fixed build date when
using SOURCE_DATA_EPOCH for reproducible builds whereas
users don't need to do anything when BUILD_INFO is not defined.

BUILD_INFO being not defined is easy/convenient for users
creating reproducible builds by default. This is also reflected by
the "BUILD_INFO ?= n" line.

Regards,
Anup