[Reproducible-builds] enable build path variation when testing unstable & experimental?

[HW42]

Chris Lamb:
> Holger wrote:
> 
>> I think regarding Debian releases, I'd recommend that we aim for
>> partially reproducible packages given known build paths
> 
> I agree. We should, regrettably, define a fixed build path for stretch
> at this point in the release cycle.

I want to note that this does not require any direct changes (when dpkg
in stretch has .buildinfo support). The build path is going to be
recorded in the .builinfo anyway and sbuild already
has a --build-path option.

So this is only relevant as a) internal definition b) how we are going
to test packages from the stretch archive.

> I do not believe we would have enough resources to fix the issues that
> would arise in time to make the case for a reproducible stretch as it
> would, of course, need to be made well before the release date itself.
> 
>> shall we wait til testing/i386 has been built once fully with build
>> path variation so we *know* the number I currently guesstimate as 75-80%??

I don't care either way.

> Whilst I'm curious, knowing the number makes no difference to me; I
> would recommend the same course of action whether it was "too low" or
> "fine". This also avoids a subjective debate on which these terms
> would apply to the number.
> 
>> shall we enable build path variation for unstable and experimental
>> for all tested archs now?
> 
> Yes, please go ahead as this reflects our goals for stretch+1. I vehemently
> believe our numbers should match our goals as close as reasonably possible.

+1

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 825 bytes
Desc: OpenPGP digital signature
URL: <http://lists.alioth.debian.org/pipermail/reproducible-builds/attachments/20160819/5cb295b0/attachment.sig>