[Olsr-dev] Repository conversion finished

[Jeremy Lakeman]

Sure if work is done first on the master branch, you can cherry-pick
it back to release.

But if you build a patch specifically for a release branch, you could
merge the fix into master. That way you can easily see from the log
that there is no work missing from the master branch that could cause
a regression in the next release.

Either way we need to be consistent.

On Thu, Aug 16, 2012 at 4:20 AM, Ferry Huberts <(spam-protected)> wrote:
>
>
> On 15-08-12 20:22, John Hay wrote:
>>
>> On Wed, Aug 15, 2012 at 08:19:02PM +0200, Henning Rogge wrote:
>>>
>>> On Wed, Aug 15, 2012 at 8:16 PM, John Hay <(spam-protected)> wrote:
>>>>>>
>>>>>> 1.) Commits MUST go to the master branch in the future. Thats were we
>>>>>> collect fixes and new stuff.
>>>>>
>>>>>
>>>>>
>>>>> merging between branches is explicitly forbidden
>>>>
>>>>
>>>> If you forbid this, how do you do the next one (2.) "fix stuff"? Surely
>>>> you fix it first in the master branch? So then the fix is already under
>>>> version control and your tool (git) will make less mistakes than humans.
>>>> :-)
>>>
>>>
>>> We will cherry-pick them to the release-branch.
>>>
>>>> In FreeBSD for instance, you are not supposed to commit directly on a
>>>> branch. You commit to head (master) and after some settling time and
>>>> if it is deemed necessary, it may be merged.
>>>
>>>
>>> The problem is that a merge always contains ALL commits of the branch
>>> that is merged. But we don't want new features, we only want the
>>> fixes.
>>
>>
>> Is that a git limitation? If so, I can understand the reasoning. On
>> others you can merge just specific commits.
>
>
>
>
> Not true.
>
> You merge 'a point in time', and you point at it through the commit. the
> stuff you merge is 'the point in time' which is _everything_ up to that
> point.
>
> I know of no tool that does this differently.
> If they claim to do, then it is not a merge but a cherry-pick (of one or
> more commits)
>
>
>>
>> John
>>
>
> --
> Ferry Huberts
>
>
> --
> Olsr-dev mailing list
> (spam-protected)
> https://lists.olsr.org/mailman/listinfo/olsr-dev