[Drbd-dev] [PATCH] [RFC] drbd: do not record build details

[Drbd-dev] [PATCH] [RFC] drbd: do not record build details

[Bernhard M. Wiedemann]

such as user, hostname and build date
to allow reproducible builds

See https://reproducible-builds.org/ for why this matters.

---
Note: There are plenty alternatives to this patch
but this one was the easiest to write for me
to have a fully reproducible build
and is meant as a base for discussing how to fix this part.
E.g. it could involve a [ -z "${SOURCE_DATE_EPOCH" ] || ...
---
 drbd/Kbuild | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drbd/Kbuild b/drbd/Kbuild
index 2d1b4d0..08def4e 100644
--- a/drbd/Kbuild
+++ b/drbd/Kbuild
@@ -134,5 +134,5 @@ $(obj)/drbd_buildtag.c: $(filter-out drbd_buildtag.o,$(drbd-y))
 		echo >&2 "Your DRBD source tree is broken. Unpack again.";      \
 		exit 1;								\
 	fi ;									\
-	echo -e "\t\t\" build by $$USER@$$HOSTNAME, `date "+%F %T"`\";\n}";	\
+	echo -e "\t\t\" reproducible build\";\n}";	\
 	mv -f $@.new $@
-- 
2.12.3

Re: [Drbd-dev] [PATCH] [RFC] drbd: do not record build details

[Roland Kammerer]

On Mon, Jul 17, 2017 at 09:49:58PM +0200, Bernhard M. Wiedemann wrote:
> such as user, hostname and build date
> to allow reproducible builds
> 
> See https://reproducible-builds.org/ for why this matters.
> 
> ---
> Note: There are plenty alternatives to this patch
> but this one was the easiest to write for me
> to have a fully reproducible build
> and is meant as a base for discussing how to fix this part.
> E.g. it could involve a [ -z "${SOURCE_DATE_EPOCH" ] || ...
> ---
>  drbd/Kbuild | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drbd/Kbuild b/drbd/Kbuild
> index 2d1b4d0..08def4e 100644
> --- a/drbd/Kbuild
> +++ b/drbd/Kbuild
> @@ -134,5 +134,5 @@ $(obj)/drbd_buildtag.c: $(filter-out drbd_buildtag.o,$(drbd-y))
>  		echo >&2 "Your DRBD source tree is broken. Unpack again.";      \
>  		exit 1;								\
>  	fi ;									\
> -	echo -e "\t\t\" build by $$USER@$$HOSTNAME, `date "+%F %T"`\";\n}";	\
> +	echo -e "\t\t\" reproducible build\";\n}";	\
>  	mv -f $@.new $@

Okay, we are talking about oot drbd, right?

I'm a big fan of reproducible builds, but only as long as it does not
break very useful information for us when we actually look at customer
installations. In that case it IMO does break that.

Again, I totally understand what the reproducible build project tries to
achieve, and it is important, but why isn't that just a vendor-patch for
the distribution you care about like the Debian guys do it?

Regards, rck

Re: [Drbd-dev] [PATCH] [RFC] drbd: do not record build details

[Bernhard M. Wiedemann]

[-- Attachment #1.1: Type: text/plain, Size: 1017 bytes --]

On 2017-07-18 09:25, Roland Kammerer wrote:
> 
> I'm a big fan of reproducible builds, but only as long as it does not
> break very useful information for us when we actually look at customer
> installations. In that case it IMO does break that.

it was only meant as an example (see comment in patch).
We can make something nicer.

https://lists.samba.org/archive/samba-technical/2017-July/121677.html
actually preferred to completely drop such build-time data collection
over my initial patch of making it optional.

> Again, I totally understand what the reproducible build project tries to
> achieve, and it is important, but why isn't that just a vendor-patch for
> the distribution you care about like the Debian guys do it?

because it is also us (open)SUSE guys, the Fedora guys, the FreeBSD guys
and there are probably a dozen other distributions out there, which is
why I try to upstream such patches as much as possible to only have to
fix it once for everyone.

Ciao
Bernhard M.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [Drbd-dev] [PATCH] [RFC] drbd: do not record build details

[Roland Kammerer]

On Tue, Jul 18, 2017 at 10:08:16AM +0200, Bernhard M. Wiedemann wrote:
> On 2017-07-18 09:25, Roland Kammerer wrote:
> > 
> > I'm a big fan of reproducible builds, but only as long as it does not
> > break very useful information for us when we actually look at customer
> > installations. In that case it IMO does break that.
> 
> it was only meant as an example (see comment in patch).
> We can make something nicer.
> 
> https://lists.samba.org/archive/samba-technical/2017-July/121677.html
> actually preferred to completely drop such build-time data collection
> over my initial patch of making it optional.
> 
> > Again, I totally understand what the reproducible build project tries to
> > achieve, and it is important, but why isn't that just a vendor-patch for
> > the distribution you care about like the Debian guys do it?
> 
> because it is also us (open)SUSE guys, the Fedora guys, the FreeBSD guys
> and there are probably a dozen other distributions out there, which is
> why I try to upstream such patches as much as possible to only have to
> fix it once for everyone.

Yes, minimizing the required patch set obviously makes sense. I will
take the first one (the sort thing), because if it helps why not.

The second one is not going to happen, sorry. Here we intentionally do
not want anything to be reproducible. We want to have that information,
that is why it is there in the first place, that was no
coincidence/mistake.

Regards, rck