[meta-raspberrypi] RFC on choice of tool for patch review

[meta-raspberrypi] RFC on choice of tool for patch review

[Alex J Lennon]

We've been having a discussion on a way forward to manage patches and
code review and would like to open this up for discussion to agree a way
forward.

Options appear to be github, gerrit, or bitbucket although there may be
others. This is in addition to continuing to send patches to the mailing
list.

Quote from Andrei,

"We dropped gerrit because at that time google dropped the support for
loging in with their accounts and gerrit didn't support OAUTH. The only
options left were involving me maintaining users / groups / permissions
etc - which obviously didn't have the time for. So, at that time, we
decided to use mailing list as the only way of patches review.

Now, I work with github, bitbucket and gerrit and I definitely, as Alex
said, feel the need of reviewing patches using a tool like these. But I
want to state the fact that, even if we decide using them, we will still
need to send patches to mailing list too - so we can keep the awareness
of this project. In terms of preference, I don't really have one. The
easiest would be github/bitbucket but I can invest some time in
installing gerrit back (as they now have the required support for google
accounts logins). So, I consider this is a community decision and, if a
have to vote, I would go for github."

Quote from Petter,

"About using github and similar, I'm a huge fan of gerrit [...] Gerrit
is really nice for reviewing and working closely together with similar
changesets that are ongoing.."

...

For my five euro-cents, I have used Gerrit a little and GitHub more. I
found Gerrit hard to get to grips with, but have been impressed with
GitHub.

So my own preference would be to use github as the UI and
fork/pull-req/commenting support all seems very accessible and intuitive.

Can others comment?

Thanks,

Alex

Re: [meta-raspberrypi] RFC on choice of tool for patch review

[Joshua Lock]

On Wed, 2015-08-12 at 10:17 +0100, Alex J Lennon wrote:
> We've been having a discussion on a way forward to manage patches and
> code review and would like to open this up for discussion to agree a 
> way
> forward.

There's http://patchwork.openembedded.org/. I'm not sure who maintains
it, nor if anyone uses it, but patches sent to the mailing list end up
there.

OOI who are "we" in this context? What benefits would a change bring?

Cheers,

Joshua

Re: [meta-raspberrypi] RFC on choice of tool for patch review

[Joshua Lock]

On Wed, 2015-08-12 at 10:25 +0100, Joshua Lock wrote:
> On Wed, 2015-08-12 at 10:17 +0100, Alex J Lennon wrote:
> > We've been having a discussion on a way forward to manage patches 
> > and
> > code review and would like to open this up for discussion to agree 
> > a 
> > way
> > forward.
> 
> There's http://patchwork.openembedded.org/. I'm not sure who 
> maintains
> it, nor if anyone uses it, but patches sent to the mailing list end 
> up
> there.
> 
> OOI who are "we" in this context? What benefits would a change bring?

I managed to miss both the [yocto] and [meta-raspberrypi] tags on this
mail, apologies for that. I guess that "we" are the meta-raspberrypi
maintainers.

Still, the patchwork instance may be an option? The meta-freescale
layers appear to be using it.

Cheers,

Joshua

Re: [meta-raspberrypi] RFC on choice of tool for patch review

[Alex J Lennon]

On 12/08/2015 10:45, Joshua Lock wrote:
> On Wed, 2015-08-12 at 10:25 +0100, Joshua Lock wrote:
>> On Wed, 2015-08-12 at 10:17 +0100, Alex J Lennon wrote:
>>> We've been having a discussion on a way forward to manage patches 
>>> and
>>> code review and would like to open this up for discussion to agree 
>>> a 
>>> way
>>> forward.
>> There's http://patchwork.openembedded.org/. I'm not sure who 
>> maintains
>> it, nor if anyone uses it, but patches sent to the mailing list end 
>> up
>> there.
>>
>> OOI who are "we" in this context? What benefits would a change bring?
> I managed to miss both the [yocto] and [meta-raspberrypi] tags on this
> mail, apologies for that. I guess that "we" are the meta-raspberrypi
> maintainers.

Yes :)

Andrei (maintainer) was using Gerrit but that went away for reasons
outlined in the preceding email.

Some visibility and control of the review process has been lost as a
result and so it was suggested we kick off a conversation on what
tooling to use to restore that.

> Still, the patchwork instance may be an option? The meta-freescale
> layers appear to be using it.

Many thanks!

Cheers,

Alex