Subject: Re: [boost] Permission to proceed?
From: Vicente J. Botet Escriba (vicente.botet_at_[hidden])
Date: 2013-10-13 07:51:55
Le 13/10/13 10:28, Stephen Kelly a écrit :
> On 10/12/2013 05:11 PM, Daniel James wrote:
>> On 12 October 2013 14:19, Stephen Kelly <steveire_at_[hidden]> wrote:
>>> On 10/11/2013 11:49 PM, Rene Rivera wrote:
>>>> In this case we wanted to hold off on trunk commits that had a high
>>>> probability of breaking many things while we worked on the beta. As people
>>>> needed to get test cycled as quickly as possible.
>>> You mean 'high risk', not 'high possibility'. No breakage from those
>>> patches yet :).
>> Sadly, that doesn't mean much. The test results haven't updated since
>> Thursday and a lot of libraries aren't checked regularly. For one
>> example, iostreams is mostly unmaintained, although I occasionally
>> review patches. I had a quick look at the changes, and noticed this in
>> 'boost/iostreams/chain.hpp':
>>
>> -#if !BOOST_WORKAROUND(BOOST_MSVC, < 1310)
>> +#if !BOOST_WORKAROUND(BOOST_MSVC, == 1310)
> That's part of a private/template friend workaround. Elsewhere in boost
> that workaround is applied for MSVC 7.1, so I concluded the < 1310 was
> a mistake, and it should have been a <= 1310 before.
>
> That's unrelated to the new patches though. That patch is from September.
This is the kind of changes that you need to confirm with the maintainer.
>
>> Which looks wrong to me. That was on the second screen of the changes,
>> and there are a lot more screens after that. The rest of it might be
>> fine, but it's unlikely that they'll get checked soon.
>>
>> Also, I asked you not to make these changes until after the release
>> was made here:
>>
>> http://listarchives.boost.org/Archives/boost/2013/09/206537.php
>>
>> Which you seemed to agree to.
> Sorry. I didn't see any connection between the release (on the release
> branch), and trunk. I asked about that, but didn't get useful
> information back, so I asked again for permission to proceed.
Daniel is right. During the release period, it is better to have an
stable trunk.
>
> I'll stop committing until a few days after the final release.
>
>> We should wait for a little while after
>> the release to give people a chance to respond to the release notes.
>> You also should notify library maintainers before making non-trivial
>> changes to their libraries.
> I'm doing that by writing to this list, right?
>
>
If you don't get the permission from the authors/maintainers or review
managers explicitly you shouldn't commit.
If them don't replay, contact them personally.
Best,
Vicente