What is the signoff policy for "arch=any" packages? Do we still need
one for each architecture? Or just one. If one, should it be form the
opposite architecture it was built on?


Allan

Am Sat, 28 Nov 2009 22:17:47 +1000
schrieb Allan McRae <allan@archlinux.org>:

> What is the signoff policy for "arch=any" packages? Do we still need
> one for each architecture? Or just one. If one, should it be form
> the opposite architecture it was built on?
>
> Allan
>

One per architecture should be fine. But don't flood us too much with
hard rules though stability has improved since we introduced
the signoff procedure. We should all trust each other and there's
always somebody around who can do a quick fix. If we stop breaking
stuff Arch fun would have gone...

-Andy

On 28/11/2009, Allan McRae <allan@archlinux.org> wrote:
> What is the signoff policy for "arch=any" packages? Do we still need
> one for each architecture? Or just one. If one, should it be form the
> opposite architecture it was built on?
I think one signoff for each architecture is the right thing.

--
Andrea `bash` Scarpino
Arch Linux Developer

2009/11/28, Allan McRae <allan@archlinux.org>:
> What is the signoff policy for "arch=any" packages? Do we still need one
> for each architecture? Or just one. If one, should it be form the opposite
> architecture it was built on?

For arch=any packages, I think that is enough 1 signoff only.
Is it right?


--
Arch Linux Developer
http://www.archlinux.org
http://www.archlinux.it

On Sat, Nov 28, 2009 at 6:38 AM, Giovanni Scafora
<giovanni@archlinux.org> wrote:
> 2009/11/28, Allan McRae <allan@archlinux.org>:
>> What is the signoff policy for "arch=any" packages? *Do we still need one
>> for each architecture? *Or just one. *If one, should it be form the opposite
>> architecture it was built on?
>
> For arch=any packages, I think that is enough 1 signoff only.
> Is it right?

The problem with this is when your 1 signoff comes from someone using
the same arch as you, and you put ELF files in the package without
paying attention, you lose.

I think there are multiple ways to make it easier. One is to show that
you have used namcap to verify there are no architecture-dependent
files in the package. Also let people know what architecture you built
it on; a signoff from the opposite architecture would be good enough
as others noted.

-Dan