Re: [RFC] collapse commits in branches to easy review

[Koen Kooi <k.kooi@student.utwente.nl>]

On 17-12-08 16:25, Otavio Salvador wrote:
> Koen Kooi<k.kooi@student.utwente.nl>  writes:
>
>> Big branches CANNOT be merged without proper review, but I don't think
>> anyone wants to review a few hundred commits :/
>>
>> So, what do you think about this?
>
> Big branches are usually wrong. Instead of using diffsplit I'd suggest
> that people provide the patches as patch series, as done in Linux
> kernel, or in topic branches.

That's an implementation detail, I think we are after the same thing :)

The thing that prompted me to write is email is this:

http://cgit.openembedded.net/cgit.cgi?url=openembedded/log/&h=john_lee/openmoko-merges

I tried to review a portion of it, but it is too much and doing a time 
based review ends up in going "hmm, this is wrong" and seeing it is 
fixed a few commits later. But let's not get ahead of things, no-one has 
asked for a review of that branch yet :)

I'd love to see that branch as patch series where we will have a few 
review rounds, but I don't know if John was enough time to do something 
like that, so the diffsplit suggestion was there to save him a bootload 
of time.

> If the user/devel has a server where he/she can host the branches, a
> set of small topic branches are easier to review and merge. Otherwise
> git send-email does the work and could send the patch set to mailing
> list.

I agree on that.

regards,

Koen