CHAOSS CHAOSScon EU 2019, 1 Mar 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Putting Order into CHAOSS: Metrics to Analyze Code Development | Ana & Daniel | CHAOSSconEU19

Description

Slides: https://chaoss.github.io/website/CHAOSScon/2019EU/slides/Putting-Order-into-CHAOSS.pdf

Session: Putting order into CHAOSS: metrics to analyze code development

CHAOSS Growth, Maturity and Decline working group provides several metrics definitions focused on code development. Using GrimoireLab we have put some of those definitions into action by setting up a collection of panels for tracking and visualizing specific datasets. During the talk we will present this collection of panels taking a deeper look to the metrics applied in a real use case.

A

Okay, perfect so, and today's talk, we're gonna basically put order into chaos and we are going to analyze. We show you different metrics to analyze code development.

A

So, first of all introducing and tell about things about us, my name is Anna Jimenez and right now, I'm doing marketing at Victoria, I'm, a gamer and also an animal over in my spare time. I love doing a lot of illustrations and doing learning Japanese. And what about you? Oh.

B

Yeah from there well I had the opportunity to work with oh.

C

B

I had the opportunity to work with Anna to produce some pilots in cages, understand, what's going on in the growth, maturity and decline, working good, so.

A

We want to serve with you. What motivates us to do when, during this talk? First of all, this understand with chaos. Emd group is also contribute and learn how to do it, because I came from the marketing field, with it's quite different from this world, so I had to I'm just a beginner, trying to learn how to contribute here and also put metrics into action and being able to integrate software and matrix, and with all that we got the new pilots so.

B

This is all basically self experience and we are trying to share with you our experience producing these panels and this working within caves, yeah. Okay, yes, well, we already know about cases. We had this presentation at the beginning, thanks to you, because so chaos is a cranium for community health analytics for open source software.

B

You can look for this at chaos, dot community or he had that comes last. Yes,.

A

Well, in sick inside GM, T group, what we, when we focus on trying to identify, where is my project at the project lifecycle if it's in growth, maturity or decline and basically start tried to identify with playing several metrics and being able to analyze that? What point.

B

Here is that we are not leading the community at all, so basically his shoes right there and Sean. That should be wrong. Oh yes,.

C

B

Subsidy or they want basically lead in the community and of course the work done here have been possible thanks to some other people around the here, so we get out there and Alberto to help us with some mentoring actions and so on. So it's working group together, oh.

B

A

Okay, so the equipment you in.

B

Decline down people here, please correct me: if I'm wrong. Basically, we are talking about three main focus area. So, if you remember, we have this focus area, we have the questions and then the metrics, so the three main of them are called development, community growth and issue resolution. So the work presented here today is focused on the first one: the code development.

A

Okay, so the first thing we did was we went to the contributed information and see what's what's going on, and we saw that one of the first things that we can do to contribute is trying to define questions and metrics to Hubbs even yeah this right here. So we took a look to the matrix that was already implemented and we realized that, even though I think they are already working on trying to visualize those metrics, we didn't have panels and yeah. Basically, we didn't have perished when to show those metrics.

A

So we wanted to create those panels and being able to visualize it in a standardized way. If.

B

You we have two technical committees in chaos, we have the matrix committee and we have a sober committee. So this is our work to integrate both boards inside because matrix people are defining metrics from a theoretical point of view, agnostic, technological point of view, and then we have this all right. So in this case there are some pieces of software they're in care, so we are not working in Mahara in this case, so we have or we have Craigie we have other software.

B

This is the list of metrics that we had a look at the very beginning. If you enter into the website, they say like isometrics so but well, we said well, it's it's! It's it's a starting point right! So as far as I remember or the best of my understanding right now, team D is trying to have a Jupiter notebook to define all of these metrics in a software perspective right. So this is another way to produce this, but instead of having notebooks, we are having balanced idea. So.

A

B

Are mainly focused on some of the first metrics that we have now for this. Basically, each of the metrics. We have specific question point that we have right what we saw a list of things and.

A

My first impression, when I saw all that was like what the hell this is a case. I, don't know how to start with what should I do so we follow a process, we open the ticket and then discuss about okay. How also we do this and then we started the creation process. We begin building the panels and also building that documentation, expert panels, a JSON file, explorer' documentation and, during the so famous per request to bloom or like sickles, and after that I had the review process and well. There were a lot of my co-workers.

A

Helped me doing this thanks to Miguel uncle Daniel that helped me a lot and finally, all those panels were merged and approved and yeah. It was cool process.

B

We had Alberto but as he was a tough previewer.

B

A

B

We have some panels. If you want to have this life, you can go to chaos that Peter Budaj Oh, which is oh, isn't he okay.

B

We can show you this like. If you go to chaotic, we turn the dial is where the website is and then we can go through each of the panels right.

B

So we took the balance. What we have is, basically we try to have this a small, so we have four. We have defined for a fight, widget right and then for each of the widgets. We have certain documentation on at the bottom and then, which is telling us what this is going on right. I have some okay.

B

We have in first place the number of lines of code change, which is one of the chaos team, D matrix things like well, Gary, hang. The panel is having a look at how they con the community contributes to repository in terms of modified lines right. So we have number.

A

B

We have number of removed lines and then we.

C

Have some specific charts.

B

At this one, where we can see some peaks of activity and then we can drill down and check, what's going on right there, because this is something that happened there, what happened probably to migration of files from somewhere to the other way in the indicators community, but if we click there well during Havana, we you can drill down, so we would say exactly the repositories were it's happened on all of these, so this is one of the panels.

B

We can go for the other one. So let me zoom out today.

A

Well, as you can see, it feels the same structure. It has a description at the bottom and also there is the main widget there, and this time we were measuring medium time to close in this, as we are measuring the full request. Instead of as there is not a merits stages, we just focus on okay. When is it closed and when it's open? So that's why we're measuring median time to close, and as you can see there are ranges, so you can easily customize the ranges.

A

So we say: okay, like creamiest, okay, it's alright median is okay. You may be taking too long already, it's just a disaster, but if in your community you feel like, maybe it can be different ranges. You can easily customize that and you can filter by organization or project you have them all the main numbers, ER and repositories. So it's pretty.

B

Superior under one day, things which is good.

B

This was, for instance, tan body by Alberto di is basically the number of pull requests. I've got merit at some point, and this is we have a trend which is the line.

B

And then the bars are basically, there are also basically the number of requests Mertz is increasing in the community. Then we have same stuff here, like filtering by urbanization, on the project list of repositories and some documentation that.

B

A

This is the average time to further spawns when pull request is open, so you can also see the tendons in line here in this case is decreasing.

A

There are also main numbers like big numbers. Are they made young timed and also the average times? First response to.

B

The change we are great yeah funding.

B

Best 50% of the.

A

And, let's end with so with some conclusions, what we achieve, we indeed understand the process and how Kahless works. We understand we learn well I I learned how how to contribute to the process. Also, we were able to put metrics into axiom and finally, we create the new path. So I think we succeed in to that and we did it kid and there's also further work to do or some highlights that my coworker danielle sue so.

B

We would like to enter into discussions at DMV, because this is worked on ourself, let's say now aside: I'm, not that integrated with the MD. So far of this book we are defining panel, try creating pilot, so there's still a bunch of panels to be created at some point, then we still we have this discussion about artists for metrics that we define these four panels really telling us something about growth, maturity or decline of our project. We are not that sure yet.

B

So we are still missing the pieces of saying what, if we are talking about the information and we build some final panel- is this really telling us something about the status of this of my parade? This is decline in and so on to something some ongoing discussion, definitely yeah, so it's basically or all of our experiencing to say right any.

A

C

How long did it take you to put together panels? I know nothing about whether what's involved in in defining the different panels. Oh.

A

I will say from my experience: yeah I mean. Maybe it took me like one week, oh well, well, for the whole process. I would say two weeks because I had to define and to understand what was that, for you know and yeah I will say two weeks like from the beginning, like discussing creating doing the pull request and getting accepted.

A

B

Not really so we we just produce or they punish to visualize the information. You can have a aggregated information for, let's say I, don't know. Maybe you are, you seem, carry the approve requests and something else for the code review process you can, you can go, you can build a specific panel for that, so that's that's, definitely doable and then you can split by data source.

B

So it's a matter of I would say that one of the takeaways that we have from here is that we need in first place to do a mock-up of what we want to have and when you really understand what you need to answer we already it is work. Thanks to the team, the people, then you start building the panel because there's this big amount of information, basically you can visualize whatever you want. So then.