[csw-maintainers] Our core values: providing straightforward experience to the user

Maciej (Matchek) Blizinski maciej at opencsw.org
Tue Nov 23 13:08:26 CET 2010


No dia 20 de Novembro de 2010 19:46, Philip Brown <phil at bolthole.com> escreveu:
> On Sat, Nov 20, 2010 at 6:26 AM, Maciej (Matchek) Blizinski
> <maciej at opencsw.org> wrote:
>> (Apologies, I hit "send" too early. Here's a proofread version.)
>>...
>>
>> For example, package consulting is hugely important.  Dissecting a
>> package, analyzing the contents, looking for direct and potential
>> problems, and providing feedback, is an immensely valuable activity.
>>
>>> Choosing the option of "no human release manager", is saying exactly that.
>>> (if one presumes that people are choosing that option, with the
>>> assumption that quality of packages will not suffer as a result)
>>
>> No human release manager means that there is no single person in
>> power.  It does not mean that packages aren't examined by a human.
>
> (I will point out that this is EXACTLY what Peter proposed: no-one
> other than the maintainer would directly examine them before release,

Yes and no.  No one other than the maintainer would _have_ to directly
examine packages before putting the package into unstable.  However,
the maintainer could ask another maintainer for a review of his package.

Once the package gets into experimental and/or unstable, anybody who
uses the package and notices a problem with it, can file a bug against
the package, saying: "this package does not follow the policy in this
place."  If we then use the bug statistics to gate packages between,
say, unstable and testing, or testing and stable, everyone can be a
release manager.

> How will they ever get examined by someone other than the maintainer, then?
> Please propose something that is actually practical, rather than just ideal.
> In the Real World, how will you ensure that packages are examined "by
> a human[that is not just the maintainer themselves]" before release to
> 'current'?

I think it was about a release to unstable, rather than current.
You're probably still thinking in the old model, while many people
already think in terms of staged package catalogs.

I don't think that the goal of providing high quality packages is
contradictory with the idea of human-free release process.

> What you wrote, seems to be contradictory.
> if there is no human release manager, then there is no human looking
> at packages before release.

The term "release" needs to be slightly redefined.  Instead of a
quantum leap from "does not exist in the outer world" to "released to
everyone", we would have stages that each package undergoes, at each
stage reaching a slightly wider audience.

> perhaps you meant "no SINGLE release manager", but that's not what you wrote :-)

Yes, it's about many people participating in staged releases.  I
believe that there being a single person with discretionary control
over releases is not a good thing.


More information about the maintainers mailing list