On 3/18/19 11:03 AM, akuster808 wrote:
> 
> 
> On 3/18/19 8:49 AM, Alexander Kanavin wrote:
>> If you do package version upgrades regularly in master, I’d say that you 
>> eventually learn about whether stable releases can be trusted. I wouldn’t 
>> need to do any research to say that boost shouldn’t be touched but OpenSSL 
>> is fine, and can similarly split the rest of what I maintain.
> 
> well openssl broke core and several other layers a few year back and
> there was an API change do to security issues and it was done in the
> minor dot release. So even that is not guaranteed never to happen again.

Yes, but the API change was documented in the changelog/release announcement.
So it was 'known', and could at least be dealt with -- if we have that
information as part of the review process of the upgrade.

I think we're in agreement, blinding accepting some change is not the right
answer.  Starting with an assumption and reviewing the facts [hopefully with
limited knowledge and resources] can help avoid these types of issues in most 
cases.

--Mark

>>
>> Alex
>>
>>> On 18 Mar 2019, at 16.39, Tom Rini <[email protected]> wrote:
>>>
>>>> On Mon, Mar 18, 2019 at 03:32:17PM +0000, Richard Purdie wrote:
>>>>> On Mon, 2019-03-18 at 10:46 -0400, Tom Rini wrote:
>>>>>> On Mon, Mar 18, 2019 at 02:01:37PM +0000, Burton, Ross wrote:
>>>>>> I don't really have a strong opinion on the mechanics behind
>>>>>> documenting the behaviour, but how do we decide what upstreams have
>>>>>> a
>>>>>> suitable stable release?  We don't want to trust every upstream
>>>>>> that
>>>>>> says they've a 'stable branch' until they can demonstrate that they
>>>>>> are competent at the job.
>>>>> I would be inclined to go the other way.  If a project says they're
>>>>> doing a stable release mechanism, and then, well, aren't, we should
>>>>> bring that up with the project and see what they mean by stable.  Is
>>>>> there an example here?  If so, is it boost?  And if that's also true,
>>>>> is there a second example?  If it's just boost, we could/should note
>>>>> that they don't follow the expected rules.  But I would hope the
>>>>> general case is that projects that say they do a stable branch keep
>>>>> it, well, stable.
>>>> I have no examples, I know nothing of the boost situation other than it
>>>> likely on balance shouldn't have gone in.
>>> Lets set aside boost then, as that is drawing attention away from the
>>> general question.
>>>
>>>> My main point was I wanted to re-document the "no upgrades" rule into
>>>> something which matches modern practises and what I think we as a
>>>> project need to do.
>>> I agree with this both in idea and general recommendation (I recall
>>> seeing what you mentioned from Greg elsewhere possibly and that's
>>> somewhere on my mind, with my U-Boot hat on).
>>>
>>>> In this specific thought experiment, if an upstream has a stable branch
>>>> which doesn't work for us we should educate them about the issues we
>>>> experience. Either things improve and we can use it or we can't. We
>>>> should document the ones we can use.
>>> Agreed.
>>>
>>>> Its likely even an upstream stable branch may screw up at times. In
>>>> those cases its then a question of judging on how they help recover the
>>>> situation IMO.
>>> Agreed.
>>>
>>>> What I don't want is people getting burnt once and then making bad
>>>> decisions based on that one experience. For example, if Armin should
>>>> now decide that stable branch maintenance is too stressful and go and
>>>> do something else it would be a loss to the project. I will freely
>>>> admit there are times when I feel like walking away as maintainer as
>>>> the bulk of what you get are complaints. I also recognise that is the
>>>> nature of the system, clearly I enjoy pain or something! :)
>>> Agreed on all parts.
>>>
>>> What I was hoping to ask / discuss is, should we go with "assume
>>> upstream stable branch works for us unless proven otherwise" OR "assume
>>> upstream stable branch DOES NOT work for us unless we research first".
>>> I believe Ross is saying we should do the second thing, and I am saying
>>> we should do the first thing.
>>>
>>> -- 
>>> Tom
>>> _______________________________________________
>>> Openembedded-architecture mailing list
>>> [email protected]
>>> http://lists.openembedded.org/mailman/listinfo/openembedded-architecture
>> _______________________________________________
>> Openembedded-architecture mailing list
>> [email protected]
>> http://lists.openembedded.org/mailman/listinfo/openembedded-architecture
> 
> _______________________________________________
> Openembedded-architecture mailing list
> [email protected]
> http://lists.openembedded.org/mailman/listinfo/openembedded-architecture
> 

_______________________________________________
Openembedded-architecture mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-architecture

Reply via email to