Re: [gentoo-project] Call for agenda items - Council meeting 2014-04-08

["Anthony G. Basile" <basile@opensource.dyc.edu>]

On 03/29/2014 09:49 AM, hasufell wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
>
> Anthony G. Basile:
>> On 03/29/2014 09:26 AM, hasufell wrote:
>>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512
>>>
>>> Anthony G. Basile:
>>>> On 03/27/2014 09:40 AM, Anthony G. Basile wrote:
>>>>> Hi everyone,
>>>>>
>>>>> The council will be meeing on April 8, 2014 at 1900 UTC.
>>>>> Please bring forward any agenda items you would like
>>>>> discussed.
>>>>>
>>>>> --Tony
>>>>>
>>>> Okay I'd like to add an agenda item.  A policy inspired by bug
>>>> #506034.
>>>>
>>>> Motion: "Significant changes to virtuals are to be discussed
>>>> (via mailing list or bugzilla) with all the maintainers and/or
>>>> herds which maintain packages those virtuals depend on."
>>>>
>>>> Discussion: "Like eclasses, changes to virtuals can affect all
>>>> the packages which depend on them.  Changing existing
>>>> virtuals, removing virtuals or adding new ones affect the
>>>> packages they depend on.  When such a change is proposed, all
>>>> maintainers affected need to be included in the discussion."
>>>>
>>> Sounds like any other reverse-dep. If people don't discuss
>>> non-trivial changes (to whatever part of gentoo), then that needs
>>> to be fixed, not policies introduced for every single situation
>>> that arises.
>>
>> True, I did think of that, but we do have a policy for eclasses.
>> With deeper uses of virtuals, it does not hurt to make the
>> reverse-dep issue explicit.
>>
>
> IMO, eclasses are special since they are not really versioned.
> Virtuals are regular packages, versioned and with stable and unstable
> keywords.

And yet we do have many such policies beyond eclasses.  For example, in 
profiles/base/make.conf we have:

# Env vars to expand into USE vars.  Modifying this requires prior
# discussion on gentoo-dev@gentoo.org.
USE_EXPAND=

There is precedence withing Gentoo for asking that far reaching design 
changes (eg USE_EXPAND) be discussed on the list.

>
> Unless you really want to apply this policy on _all_ reverse deps.

It is difficult to enumerate _all_.  I believe Tom and I have identified 
two more areas were we should encourage list wide discussion when the 
consequences by affect others: virtuals and profiles.

>
> And if people don't follow it, then what? There have been a hundred
> situations (including eclasses) that were sufficient to say "stop" to
> some people who regularly do undiscussed, non-trivial changes. Until
> now, not much has happened about it. I doubt that this policy will
> change any of that.

If nothing else it raises awareness that we are a community where our 
changes affect one another.  The "enforcement" I would hope for is the 
respect that we show one another.

>
> [sarcasm]
> But it gives us an impression of progress, at least... and policies
> seem to be a trend lately.
> [/sarcasm]


-- 
Anthony G. Basile, Ph. D.
Chair of Information Technology
D'Youville College
Buffalo, NY 14201
(716) 829-8197