Under http://koti.kapsi.fi/~federico/crstats/core.txt is a list of actions, but having that under control of wmf is better for updates and traffic.

In T86152#963373, @Umherirrender wrote:

having that under control of wmf is better for updates and traffic.

{{citation needed}}

All developer have the right to leave, in that case the statistic maybe is lost or needs rewrite from another user or the wmf. When the statistic is under wmf control and the main developer is gone, than there is sucessor which has to maintain the statistic. When there is no sucessor at least the code is under wmf control and maybe can be reused.

But that is only my opinion, not a necessary point to fix this bug.

Let's show a graph with the number of active users on a monthly basis. Then let's have a table with a list of active users in the past month (username in korma, affiliation). This way we can also identify current contributors missing affiliation.

We finally defined to have it in a quarter basis as far as I remember. Although it wouldn't be an issue to have it using a month period.

I modified the description of the task. The idea is to have some initial chart and list of people to discuss about.

Hi, numbers on participants in Gerrit.

A participant is defined as any developer that has left any trace of activity in Gerrit: comments, reviews, changes, patchsets submissions, etc.

First, the evolution of participants:

And then the full list of participants (2015-01-24, 2015-02-24) with their affiliations and total number of events. In this case, I didn't ignore self-merges. Those are simply actions in Gerrit I guess. That's the reason why you may see bots in the list of participants. We can remove them at some point in any case.

Code available at https://phabricator.wikimedia.org/P329

Ok, this looks good. Thank you!

Is this data already counting all repositories in Gerrit (T86154)?

The shape of the diagram looks a bit weird between Sep 14 - Jan 15 (flat lines, that depression in November... was that caused by the Phabricator migration?). In any case the type of chart needed is that one.

Yes, the bots and their impact should be removed from all the metrics in Korma. I hope their "unknown" affiliation is not affecting numbers anywhere.

It is very useful to see how many Unknown for so many known names. Looking forward to know how you want us to provide this data (T88277).

Thanks!

It's counting around 1200 Gerrit repositories according to the database.

That depression in November/December may be because of the Phabricator migration, but I'm afraid that I can not conclude anything. This sounds like that in any case. Just to compare, the current number in February (not added to the chart) is similar to that month, even with a one week left. So, it's pretty low.

Regarding to the flat lines, that means that we got similar or the same numbers in two different months. This happens sometimes, although I agree that it looks weird. (I double checked the dataset in any case).

Finally, regarding to the affiliations, we may move those to "special" affiliations such as "Bots". This may help to avoid noise in other charts taking into account the "Unknown" affiliation.

Just to clarify, it's counting all of the repos according to T86154: Remove the filter for key Wikimedia software projects in korma.wmflabs.org.

And regarding to T88277: Instructions to update user data in korma, we still have to discuss about that internally. In any case, a simple way to provide this data seems to have that under a revision system. This would be a list of people and affiliations that you can change and we later import.

Self-merges must be removed.

Why? Here we are counting active users in Gerrit. If a user logs in to Gerrit and self-merges a patch, that user has been active that month.

Why?

Self-merges are not code review. The bug summary says "code review". If you're not interested in code review, please change the summary.

Here we are counting active users in Gerrit.

That's not what the summary and task description say.

If a user logs in to Gerrit and self-merges a patch, that user has been active that month.

That would require sending the patch first, so it's best identified by counting patchset/changeset submission. If you're interested in any kind of activity and in mere counts of people, then why attach numbers to names and why focus on CR? You may consider patchsets, comments and CR and remove the numbers.

As similarly done in T59038: Basic metrics about contributors exercising +2/-2 permissions in Gerrit, the following step consists of adding the proposed chart (nicer and with addition of dates, etc) and table to the Gerrit overview in Korma at http://korma.wmflabs.org/browser/scr.html.

For those interested in the mockup discussion, we're having it at T59038: Basic metrics about contributors exercising +2/-2 permissions in Gerrit.

A basic data point that we are missing at http://korma.wmflabs.org/browser/scr.html:
How many users perform any kind of action in gerrit.wikimedia.org on a monthly basis?

Errm, isn't that "Participants" on http://korma.wmflabs.org/browser/scr-contributors.html ?

@Aklapper, that panel is work in progress of this task.

We're still missing two points in the panel to close this task:

• Removal of bots
• Addition of affiliations.

This panel is ready to go.

Please have a look at it.

Affiliations were added and bots removed as final steps

http://korma.wmflabs.org/browser/scr-contributors.html

This looks good. Thank you! We are still discussing about labels at T59038#1231001, but that should not block this task.

Innocent question: have you checked the data well? The number of monthly users in Gerrit since Feb 2014 is pretty stable, which is kind of shocking considering the growth of the engineering departments at the WMF...

In relation to

In T55489#1233269, @Nemo_bis wrote:

Why is there no "unaffiliated"? All these graphs give a very partial picture.
http://korma.wmflabs.org/browser/scr-companies.html

and

In T86152#1227275, @Dicortazar wrote:

Affiliations were added and bots removed as final steps

Can it be that we lost "Independent" and "Unknown" with the change of system to introduce user data? We had these values before. Now the tables in http://korma.wmflabs.org/browser/scr-contributors.html only show a dash, and not having a single identified "Independent" among the Gerrit users of last month shows that something has gone wrong.

Thank you @Nemo_bis, I had missed this detail when checking these tables yesterday.

Ok, we still have such information, but as you mentioned, this is not considered in the analysis.

Please, let me check. And thanks for the pointer!