Re: pkgsrc-2013Q4 freeze started

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

dholland-pkgtech@ wrote:

> On Fri, Dec 20, 2013 at 09:04:54PM +0900, Izumi Tsutsui wrote:
>  >>> My suggestion is "make the rule more specific to avoid post-commit 
> flame."
>  >> 
>  >> My suggestion is "think before committing, and if in doubt, ask pmc".
>  > 
>  > There was no doubt that updating leaf packages didn't require
>  > prior approval per the latest announcement:
>  > http://mail-index.netbsd.org/tech-pkg/2013/03/17/msg010920.html
>  > so the complaint against MAINTAINER's update looked just bikeshed.
>  > 
>  > > It's worked fine for me for a number of years now
>  > 
>  > It failed recently for others.
> 
> The most fundamental rule is: don't break the stable branch. If you
> can't see that a last-minute unannounced/undiscussed update to firefox
> would cause people to be concerned that the stable branch might be
> broken...

The update was done not last-minute (last ~10 days).
Leaving older firefox with known security problems in stable branch
looks rather embarrasing.

You claimed "firefox was highly visible important"
but you have never defined which packages are "important".
agc also claimed it was deteremined by "common sense"
and I claimed it would not work, but no answer.

My suggestion is "define it more specific to avoid bikeshed"
as noted the above.

> I'm not sure I can explain any better than I already have.

No principle explanation is needed. Just define
which leaf packages you want prior approval before commit.
Otherwise fewer developers will update packages during freeze
to avoid bikeshed.

If you want no-breakage during freeze, you can simply define
"no leaf package updates are allowed" or
"all leaf package updates also require approval."
Why not?

---
Izumi Tsutsui