Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/confuse: confuse-2.7.ebuild ChangeLog

[Michael Orlitzky]

On 03/10/2013 02:11 PM, hasufell wrote:
> On 03/10/2013 07:04 PM, Jeroen Roovers wrote:
>> On Sun, 3 Mar 2013 12:44:18 +0100
>> Tomáš Chvátal <tomas.chvatal@×××××.com> wrote:
>>
>>> If I remember correctly the damn rule is to put it for 30 days into
>>> testing, and as you said there was no previous version on arm so users
>>> could've reported some issues, i agree that sometimes you have to
>>> ignore the rules to really fix stable, but was this such case for
>>> sure?
>>
>> I've done straight to stable keywording _many_ times. The rationale is
>> that with no previous version stable for a particular architecture,
>> there really are no users who could see _regressions_, hence waiting
>> the nominal thirty days is meaningless in this case.
>>
>>
>>      jer
>>
>
> another note:
> I was told a while back (I might still have it in irc logs), that 30
> days is NOT a rule. It's common sense, but in the end the maintainer
> decides when to request stabilization, no one else.
>
> Blame people if they break something, not if they ignore soft policies.
>

What's broken is the expectation that the package was tested by more
than one person. The "soft policy" is here:

http://devmanual.gentoo.org/keywording/index.html#moving-from-~arch-to-arch

And you're right, ~30 days is simply a suggestion. But the rule is "The
package has spent a reasonable amount of time in ~arch first." A further
non-suggestion is "The package must be widely tested."

If a package is marked stable, I expect it to have seen some testing,
and not just on the maintainers personal machine. I don't rely 100% on
the stable designation, but it does affect the amount of testing that I
personally will do. Please help me do my job by not perverting the
meaning of stable.