Contributors mailing list archives

Re: Proposal for new workflow, incorporating "Optimistic Merging"

redCOR AG, Robert Rottermann
- 07/06/2016 20:17:20
Hi there

On 07/06/2016 19:08, Nhomar Hernandez wrote:
<blockquote cite="" type="cite">

On Tue, Jun 7, 2016 at 11:53 AM, robert rottermann <> wrote:

I believe we should try to find a way along this lines.
The actual situation is plain bad I think.
It actively discourages contributers and blocks good stuff waiting to be evaluated
EVERYONE of us should be able to check the code.

But EVERYONE needs a handle to judge what code is good and stable.

Hi Robert.

Thanks for the comment.
de nada
<blockquote cite="" type="cite">

Just FYI, now we (the commiters) only wait for 2 +1 (it is the rule) and EVERYONE can give a +1 in a PR, for responsability we ask make a real review but honestly that's based on trust some people say "+1 functional" or "+1 technical but I think x,y,z".... o not like it ..
It's not the way I work.
Or better said, this release workflow does not easily become a natural part of my development workflow.
If I need something I hunt for it, download and start to use it. Very often there is a need to interface with what I downloaded in some unique way. Thats when I start to learn its innards and stumble over problems. I probably improve little things ..

Nothing new here.  Thats how all of us work.

But what do I do with my findings (if any)?

I want to push it to the repo, so that the rest of the world (and me in some other setting) can easily use it.
I would not do it lightly, I might ask somebody to check it when it is more than cosmetics. I would make it pass all tests..
But I want it done and out of my mind in a very short time frame.

If this is not possible i fork and make it part of my own tool set...

<blockquote cite="" type="cite">
We are more like commit channels (like the high level team side by side Linus Torvalds) even he say that honestly when he mention "I trust in no more than 10 persons, if they tested I merge" well that's more or less the same here.

You can easily Pick any PR (OCa member or not) just test it (functionally in runbot no technical skill needed) and give your +1.
As said,  I want to use and fix (if necessary) I do not want to give any stupid +ses.
I hate all this millions of messages with nothing but a +.
If a tool I use would be marked with "please give feedback on .." I would do so
<blockquote cite="" type="cite">

Then the fact that EVERYONE can judge is a matter of fact today, you +1 is respected and if you considered you tested partially something, just mention that than another person can give a +1 to the part you did not test.
The only way I can test is using, adapt give back..
Of course, there should be well defined rules that control the process on how to change things. But they should allow that 90% of the pull request are resolved within a very short time.
<blockquote cite="" type="cite">

I hope see you in github I am sure you will give .good visual of what you use, at least to giv feedback.

I think we must prepara a video with this working becuse the impression we are giving today is that we the mergers has the power of Block which is not totally true, we have the responsability of merge which is different.

Why no more Open, it means 100% have merge power?

We had that in this repository years ago:

And the results were really BAD, the process we have today was to enable a minimal quality check because nobody was responsible at that time.
This I can easily believe. No RELEASE should be done without testing and approval.
But there should be a development repository where every OCA accredited developer is allowed to push.
And it should be sensibly easy to become an accredited developer.

<blockquote cite="" type="cite">

Now you can know who is in chage of what, and you can (by yourself) simply give a ping if something is blocked and give you own +1 (even functionally).

I love the process we have now,
because you are NOT one of the many developers that are blocked ..