2010/8/12 Mario Figueiredo <mario.figueiredo@quiettech.org>
On 11-08-2010 18:03, Pierre Schmitz wrote:
On Wed, 11 Aug 2010 17:41:06 +0100, Mario Figueiredo <mario.figueiredo@quiettech.org> wrote:
This would definitely get me interested in Testing. Right now my Linux knowledge is limited and thus Testing is a no-go zone. If however I could have a guarantee that Testing offers the same package sanity insurance of the other mirrors, I could start participating.
In that case testing wont still be for you. There wont be any guarantee for testing and some pacakges might be just broken. The only thing you can expect that we wont break testing _by intention_ due to moving incomplete rebuilds in.
Well, that was precisely my point, wasn't it? Testing implies bugged application builds. What it should however not imply is broken packages.
[testing] implies bugged application builds /and/ broken packages. That's the point of [testing] to identify broken packages before they go to [core].
It needs to be said that this is also reflection of what one should expect to encounter in the development process in the wild. Apart from the potential for collaboration, the idea that the Arch repos could mimic this development cycle is very appealing to me.
__________________________ | | V V Development<-> Staging<-> Testing -> Release
Packaging maintenance is taken away from the end user, giving them "safe" (it's still a beta, hence the quotes) access to Testing. Meanwhile developers would separate packaging from Testing, considerably giving them a lot more control over what users can access from Testing.
Staging is not a new repo/layer between the developer and testing. It's just meant to be a temporary storage for rebuilds. The current dev. cycle wont be affected. So we'll still have: dev->extra dev->testing->core
Aren't you contradicting yourself? Unless you don't plan to use staging, you won't risk anymore having broken rebuilds on testing.
No, but there can still be broken packages in [testing].