[LEDE-DEV] Proposal to sign all commits

[Daniel Dickinson]

On 16-05-06 02:58 PM, David Lang wrote:
> On Fri, 6 May 2016, Kus wrote:
> 
>>
>> I'd argue such a barrier is OK if we want to quickly increase the size
>> of the team of people with commit access. I think we're
>> underestimating our contributors here. I agree that we shouldn't have
>> unnecessary barriers (such as copyright assignment to give a specific
>> example).
> 
> I'm not representing LEDE, but I have to ask, why do we think that we
> need a lot of people with commit access?
> 
> Linux Kernel works just fine with only one person having commit access
> (Linus)
> 
> I think this is legacy thinking based on SVN.
> 
> With Git, you don't need a lot of people with commit access, everyone
> can have their own tree and create branches to be pulled by a single
> person (or small group of people), potentially through a hierarchy
> similar to how the Linux kernel maintainers merge things from their area
> and then Linus merges from them to the final tree.
> 
> Too many people with commit access to the central tree just leads to
> more ways to trip over each other.

I agree with David, enough committers to prevent merging being a
bottleneck, but no more than actually required for merging (which if the
trees being merged are in a respectably up-to-date state shouldn't
result in huge numbers of conflicts that need to be resolved, in fact
I'd argue that tree really should already be up-to-date vs master so
that you only need to do a fast-forward update before merging, as
untested merges create the potential for untested changes that cause
issues (merge is just textual not semantic, which means that even a
successful merge with no conflicts may in fact break things)).

Regards,

Daniel