AsyncAPI Community, 21 Mar 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Community Meeting (March 21th 2023)

Description

https://github.com/asyncapi/community/issues/640

A

And we are live, hey everyone! um Welcome to today's community meeting I'm Haze um you're, really glad that your regular guy. um So to be honest, I don't have like much on the agenda today. um But if you have like any discussion, you want to bring up to make this meeting long.

A

um I'm always up um so I. Basically just have like I think one or two things on my agenda to um talk about and the first one is actually from the um past um mentees of this mentorship program. uh So I keep getting um questions. That's do we get excited certificates for the Sync API mentorship program and I was like I I, don't know, but I decided to like bring the discussion of today.

A

um Yeah I think my question should go to do you think we should have a certificates that mentees can showcase, maybe on their LinkedIn Channel.

A

Just like, because I'm not good who.

B

Would sign them because.

C

B

Were talking about like official certificates or um because like we can have some badges, um we had some budgets anyway, invented by Missy, so which was a matter of extending them and putting the name of the of the person that completed the mentorship. That's doable certificates. It really depends what they want, because certificates from hyper stack objective. It means already something super serious and official and certificate for me is something that somebody signs and like with the name and blood um so not like what they really mean by certificates.

B

Maybe it's enough that we produce this nice, basically banners that they can share on social media, foreign official banners that we create through GitHub and then share with them. So they can. You know like.

A

I think I think there was something like Google some of gold.

B

I'm not sure what Google summer of code is doing. Oh did you get something yeah.

A

Yeah I I didn't get a certificate.

B

What what did it say because, like.

D

Like honestly,.

B

I have no idea what like, so, you would have to show me something like so I can actually understand what people really need. Yeah.

A

So, basically, uh the certificates, as far as I can remember the certificate contained um the name of the organization I participated in the projectile participate in um official signature from Google uh to show like yeah I like him, yeah written, okay, I, don't think it's n, like e-signature I, guess.

B

Because, like legally I, don't know who could sign it, uh I mean like now: I I know that you're.

A

Probably right.

D

C

Director but like.

B

It's not the director you can imagine from a company like I'm just here to serve people not really sign papers.

A

B

Signature has no value, that's what I mean. Basically, um um so that's tricky. One um is signature, a must-have.

A

um Well to me: I I, don't think it's a mouse. We have signature.

B

Because, like I said like we can, um we could work with someone ask to extend this, this official menti logo, with some generated paper, where we say, like uh congratulations like certificate, congratulate for completing the mentorship, uh 2022 and working on this and that project with that Mentor. Maybe even uh we can, we can do it. It's just a matter of like getting some design simple design in place that we can generate. The critical part is just this.

B

A

B

People are like because I I guess that people probably want to have signature, so it's not fake but come on like it's. Anyone can like it like faking signature, it's pretty easy, I guess.

A

So this is wait. No you already make these like more complicated than I was thinking because I was like certificate. Of course you can do it just put together some stuff and get your cash like sign it for now, I.

B

Mean so I I know what white people ask for it and it's pretty cool that they can like um publicly not only state that they completed the mentorship but actually has some sticker. Whatever uh I know we promised swags, but that's uh something that is delayed and we're gonna, deliver them and certificate.

B

Certificate I mean we can even have a.

B

C

B

So we basically have to have discussion because we have to involve a few people. We could, for example, check with barbano and um okay, because barbano she's, like she's the one that is working on the swag store and- um and we will have like the plan- is that this wax store will have a item called mentee package, whatever uh that you can like, uh so the menti will get a a a voucher from us and they can claim the package, so nobody can buy it.

B

It can be only claimed by the voucher and and then the company that has Warehouse with the with the items they will send it uh to the mentee.

B

um And uh and uh and- and maybe barbano can check with this swag company, if they could, together with the uh with the with the hoodie, also send a certificate, we will just provide them with the template and maybe in this swag store, we could have additional Fields, where the Mandy would specify the name and the project that they were. They were working on, so that could be done.

B

And then we have this super strong assurance that it's um that it's not fake, that it's, that people can have such certificate printed out and handed over only if they really completed, because only people that completed got the voucher.

E

You could also have a link for the issue itself that actually States, who are have done it yeah.

B

For example, so that can be done obviously like we didn't promise anything like this for last year, so we would probably have to find out simply just some work around, um but yeah I mean, let's do it if mentee is needed, let's do it, okay, whatever we can for them.

A

B

We can hire someone to make the signature.

A

Really you can hire someone to sign, there's.

B

A you have a template and you have to sign like in the template.

A

Oh, my God I can't believe he's saying this I'm.

E

Pretty sure you know how to code as well, because so.

B

E

Just an API that can do as well.

A

Okay, okay, so I'm going to start a discussion around this um unless, like other people jump on it,.

B

um We can also ask chat: GPT, maybe can sign it for us.

A

Now, I'm panicking.

A

Okay, um so the second thing on my agenda: is um they useful interest in stuff? That's like have like a lot of questions in it, so I'm going to share my screen.

A

um Do we have it? Okay, very nice, Square.

E

That's quite a few Taps.

A

Yeah um against my screen.

C

A

Okay, cool, so is it bad enough or should I.

E

Or is it yeah, maybe so many other than.

A

Okay, so, um as you all know, we are actually uh trying to like Implement um a contributors page in the new community. Section um we've been working on um so the whole idea of these um contribution. Contributor selection is to have um a page dedicated to the SDK by contributors, which is really nice uh for me and.

A

Analysis of the all um contributor stuff and how much they've contributed, how much issues they've opened and yeah. So this is the result right. So this is the results um this. This is not the total results like so we have two results, so I just wanted to like have your inputs and see how we can go so we have to resolve.

A

D

Yeah so uh see there are some projects that uh we own right, like we are the code owners, so we are not necessarily having pull requests to go to that, sometimes to directly push it. Some changes right uh so, for example, the problem Repository so I'm maintaining that so here we should have a column separately, for if there is any project which is owned by the particular GitHub handle, we can mention that, for example, sawbik has cliently here is the code owners there? So all all these also counts in the contribution right.

A

Sorry, you push directly to Mazda.

D

No, not the master I'm saying like we have our own separate, Branch right. uh Suppose I have created my own Branch, so a b development and then I create one pull request. After having set of changes done I hope you are getting my point. Yeah.

A

It's gonna count.

D

That will be counted as 1 only right, yeah.

A

D

There are a lot of changes, suppose 10 10 to 15 commits in that branch. I.

E

Think it's important to maybe declare what a contribution is so generally I think that the general consensus is that a contribution is a change to master or to the main branch or to some of the sub branches that are public.

E

So that means that your changes are taking effect into the code base and that's counters one, no matter how many changes you basically do.

E

Those counters won the same with issues like new issues, just a new issue, but how many things you do on your local Brands and your local Fork doesn't count as a contribution.

E

If that makes sense,.

D

Did it make sense as well I think it was like we have only these two parameters for what about like uh the maintainers, who are constantly seeing different uh things right in the project, the project they maintain texting? The number of issues uh discussing with the other colleagues.

B

Yeah so so, basically um there are contributions like comments, reviews.

D

C

B

Are the things that um abir talks about like um that? Like you like one thing?

B

Is that, like it's cool that you created a pull request, but this request requires a review of several different um actors um and that's also a contribution that is super hard to measure and like, for example, the Bounty program, we're thinking about it's um that cannot cover it um because that's pure maintenance work that is super hard to measure and and the same with issues like uh sometimes um it's like um the the issues created and somebody can be accounted for creation the issue.

B

But then it's um there's a discussion involved, um a lot of comments and again like um maintainers, and some other contributors also are involved. Asking questions you know like. um So that's what it's missing basically uh from the list, unless, unless issue status, is something different than just issue creation.

A

Yeah, like so I I, basically understand so. If we are going to like Factor all these concerns in then it would be very difficult to measure and I. Don't think like a GitHub archive like like covers all the stuff.

B

It must cover it uh because it's all about its comments right, it's uh like GitHub archive, is like it holds all the events, basically as I imagine GitHub archive, but I. It's basically every single event that happens on GitHub goes to GitHub archive. So it's like. There are different types of issues. Then it's a issue. It's like with GitHub actions like you have GitHub action that reacts on on merge or on pull, request, create or review uh comment create or comment create.

B

So there are um even for sure events that um that are recorded that contain information about the comment that was created and author of the comment not.

D

B

Of the issue but of the comment, so this additional stuff should be and and in theory, is in GitHub archive all the way, don't know if it's there, but should be.

D

That's how like we should figure out like different criteria.

A

Okay, so um we we have sorry, we have actually um so I'm, not I'm, no expert in this data stuff, um so I'm pretty sure. Tony should be able to give us some insights because she eventually like walked on the whole, the other stuff, so I, because I I have a concern. Also because when I say issues right, I was so skeptical about the amount of issues I'm seeing there because I saw I was feeling okay.

A

This can't be the amount of issues I've opened because to me I'm, not sure if I've opened up to 10., but in this time frame right so I was thinking. Maybe it started, adding my interaction with other issues as part of contribution.

B

Yeah might be that it's already including comment, but.

A

But now I have a concern, so let's say for instance, um right and I just come there and I comment: hey I'm interested in this issue and I leave. So it's basically going to count down as contribution and basically.

D

A

Can contribute.

D

No, it won't be counted as see uh the comments.

A

Yeah, it's real.

D

uh Like anyone can do that right, you have to have a separate distinction. The comments in the pull request and the comments in the issue.

A

No, no I don't have up to 10 issues.

D

So I'm asking that uh see when you comment on the political, it might be a review comment or something like around it right. So there's a bandwidth, consumed and uh there's ownership involved there, okay or the one who is maintaining, but in the normal issues uh people drop their ideas and all these things that's perfectly fine. We are not explicitly saying that okay comment over here and all this right, but in pull request we have to That explicit approval before merging it, so they are. The consumption of bandwidth.

D

Is there so we have to count that as a contribution instead of issues I. That's that's my point of view I'd like to hear from Lucas as well. What are things around it.

D

B

Also, my contribution right um I mean sometimes it can be just hey, I want to work on it, and sometimes it can be a pretty extensive comment with that adds much more value.

B

um That's something you can't like without chat GPT, you can't obviously.

D

um Over there, then, we might have to do a sentiment analysis of the comment there, whether it's adding a value or not.

B

Analysis is for and I would definitely not recommend sentiment analysis because we're.

D

B

It already for for some time and it's it basically sucks, because the sentiment analysis doesn't have the context so sentiment. Analysis can tell you that short comment is um like it's still. It will just give you a sentiment, but still it it uh it doesn't it's not able to tell um the import the value he doesn't know the other, like the rest of the conversation. So still the sentiment, uh it's not really of any value.

D

um Okay, let's count the comments then as well. That will be fine. Nobody.

C

D

Going to produce too twice right, it will be only once you know, issue Ace is not going to introduced in multiple issues right. So that's one thing.

E

One thing I just want to mention is that this is not exclusively related to async, API or open source on this specific uh problem, like this is performance related that any company have problem defining like how do you? How can you measure performance of a developer, for example? No, because there's things you can't measure, that's the reality, so it the important part is that no even with jira it doesn't matter. Do you have interpersonal stuff?

E

Maybe the sidemen that you have on yours like if you're sitting next to someone, maybe they contribute by a pair programming that isn't seen in a gr issue like there's so many things that you can't take into account, and this is a problem in with development in general.

E

So it's not a native problem for this I just want to mention that maybe some someone has figured this out in general that we can take notice of and use as guidance, but I also want to mention that if this is related, what's this related to the um Bounty program stuff as well right or this is completely unrelated. Okay, it's.

B

Actually, more related to a a view that we would like to have in the website in the future, where we have all the contributions, all the contributors gotcha but yeah as but I agree what you say like it's. um It's a it's, not a new problem. That's somehow popped up! It's a it's, an old issue: how to get numbers to measure contributions.

B

um Numbers should never be treated one to one so they're, just supporting and- and there can be like so so from my experience, I can tell you like Linux Foundation provides us with some basic um GitHub stats um and when I was like recently when I was publishing this blog post about the summary of last last year and I used to the there um and I use their data uh in the in the article and um the diagram was clear that it's um showing the country all the different contributions.

B

So it's not only code, it's it was explicitly in the um no don't remember exactly where, but it was written like it's a com. It's a review comments. It's uh interactions in the issues. Everything is counted because, like it's really hard to say, if again like, if a comment in the issue, it's uh it's useless or not. Sometimes even a comment that is like hey can I work on it. It's it, it might be useful actually, because it pings you, you jump into the issue as a maintainer, you can revamp.

B

Oh actually, that's issue is not no longer valid, I'm closing it and it's still a contribution like somebody contributed basically by um waking. You up. Oh there's an issue. So so it's still um it's delay contribution I, would say so, um but yeah in total numbers. You're gonna see like. If somebody has this one tiny issue for hey I want to work on it.

B

Still it's just one issue like it's. It's fine and and also like when you go like even when you go to your profile on GitHub, you can see this I'm, not sure. If you have it enabled you can see this graph. That shows like uh shows your code review contributions, issues or requests. So it shows how your how you really do open source. If you just comment, comment commit or you actually.

A

Collaborate exactly.

B

Yeah yeah, so that that's the same uh the same or similar approach you should take here like a um if you will have only issues and nothing else, so the graph is pretty flat and if you will generate some spam issues this and like, if somebody will try to you, know like hack the system and start doing this I want to work on this. I want to work on this in different issues.

B

um Numbers will just confirm what we will already notice on our own. Like hey, we have some spammer here that wants to just override the system, and probably it wouldn't. It will end up with blocking the GitHub user.

B

um So I would not worry much about on specific items, but more about what we measure and we should measure every GitHub activity, but just simply transparently say what number is what that's it.

A

Yeah I think I think that's perfect. What.

B

Is the most important is that we already see something that this GitHub archive Beast? That is a huge project that um that you managed to get.

A

C

To get something fantastic.

B

Oh yeah, we just we just need to like make it clear what where the numbers are coming from and that's it.

A

C

A

Is interesting and now I think the most interesting aspect of what truly did was actually um dividing the measures right, so we have like the way, get a buckeye Works. Sorry, according to our explanation, the way gilobarchive works is because the initial initially we were trying to like get data from 2016 to current right. So, but we noticed right, um it takes up. It takes some time for the data to update and um we will be having like different numbers. So what she eventually did was say.

A

Okay, now we're gonna have like overall data right and the overall data is going to be coming from um 2016 until 2023, right so from 2016 to 2023, which means just data from 2016 till the end of 2022. So that is the overall data, but the correct data is going to be coming from 1223 January 2023, the latest right you might want, you might be wondering: why is it? Can we just like, have everything together? Okay, but the way you get a bucket works is when you call data from 2026, 2020, sorry 2016 to 2023.

A

It doesn't give you the data for 2023., so it only gives you the data from 2016 till the end of 2022, so the yellow, colon or the yellow query only runs at the end of whatever year you called so, which is why we have like two separate details. So if you're calling from 2020 2016 to 2023, you are only going to get data from 2016 to the end of 2022., so it won't include data from 2023. I, don't know! Why um can you share some light.

F

Okay, um so the dates are basically installed into tables right, so each month of we have tables for each month. We have tables for each week. We have tables for each day and we have tables for each year.

F

So, if I write a SQL query that I want data from 2016 to 2023 16 to 2022, because the only tables that the yearly tables that are available is from 2016 to 2022 and 2023 table hasn't been created, so it will be created like it will like they will combine each table for each month um at the end of the year. So this is why we say for weekly data.

F

This is they're, currently dumping data and, if I say, okay, give me data from 2020 to January until present, because the data is dumped each and every day. That's the only data available, but I cannot get data like okay. Give me from here this to here this because for 2023 the tables haven't been created, I'm, not sure. If I explained it correctly,.

B

But you're saying that there's, but there is stable for January 2023 and February 2023.

F

Yes, but it's not under the yearly data, but it's only under the month. Data.

B

Okay, but um what blocks us from using the the other tables uh like can't, we just say: okay, so we grab total from 2016 to 2022 and then in case of latest so 2023. In this case, we grab it um monthly, um like monthly.

F

um So, like, frankly, we have archive data from 2016 until the end of 2022 right. So we got the data like the pre-request, the issues that a person has owned.

F

Then, since we don't have data included from this year, we are querying like for the from January until present. So this script we are running like every week because of bigquery also has limits I'm still I'm, using the free version on how much you run and also for to query data from GitHub archive you either have to query by date by month or by year. So those are the three um methods to create data from gig tab archive.

F

So if I am to query data from 2016 from this month until 2016 this year or from this year 2016 until the year, 2022 I will get my then if I am to pray from a month, uh like maybe we'll say from 2023 from month, January until present, I will get my data if I want to create, like okay, for instance, the beginning of February, on the 1st of February until the end of February I will get my data I'm, not sure. If I'm experience explaining it correctly.

A

Is asking is if it is possible for us like how this I think sort of a match right, so we have body data from 2016 to 2022 right. Then we made it with the results we get from 2023 to current. Then we sort of, like you know, match it together as one.

F

Yes, we can major together, but now, since we are updating the data frequently, how are you going to do that? Oh, do you get what.

A

I'm saying the.

F

Data is being updated frequently. This is why we have two separate tables, one that is current, like the one that we're currently running like every week and the table that we already have. That is full from 2016 to 2022., since the one for 2023. We are currently updating it like every week and git tab is currently damping it every hour or every day, I suppose so. This is why we have like two separate tables.

A

They think you're, the idea person here.

B

Yeah I'm afraid to admit I, I I'm, not 100 sure uh so so so we have data which is static from 2016 to 2020. Excuse me.

B

Excuse me uh from 20 uh 2016 to 2022., and then we have uh data from 2023.

B

That is um not I, think because every single day it's um it's refreshed and um but and we fetch it weekly and dump it to bigquery right.

B

C

F

We are running the script on bigquery, then I'm exporting the data to excel.

F

This is what you're currently seeing here.

A

So I think watch I think because it is not possible for how to like run these two data um or these two requests simultaneously together. Right.

A

So I think there's one query: I I, like I'm no expert I'm, just like saying so I think. um Let's say this one tab. So this particular tab only um queries the data from 2016 to 2022, so in that tab, I'm, not sure if it is possible to also like run the query that, like generates data from 2023 to current, you have to like do that on a separate um tab or something.

A

Oh totally am I right.

F

Groups so one for the static data and one for real occurring data. So the one.

A

F

Until 20132 is a separate script, it's and these own tables except rip them the one that we are running in, like big sorry. It has its own.

A

Three I think you're breaking.

C

A

F

A

You're explaining the scripts.

B

Yeah, so we last thing we heard was that there are two different types, basically like one for the the start again, one for her for the latest.

F

Yes, one for the reoccurring data. Yes,.

B

But like technically, we can merge it if we want.

F

Yes, technically, you can match the tables like. If you have the current data for this week, you can make it to the old data.

A

Yeah and that that is what we'll be doing like, because I'll be fetching the data from the Excel stuff right and converting it to Json. So then from there look at the conversion I'm going to like merge this to data together to like get the latest stuff, but now, yes, here's the football thought right, because, according to the discussion, I was having truly we. If we follow this approach right, we folks, like bukash, Jonah's, Fran machine me eventually. I cannot always be the people at the top of the list.

F

Yes, thank you for pointing out that you, because there are people who contributed like in the middle, let's say 2018 for instance, or maybe people have contributed this year. I mean like Lucas is like how many I think of a thousand ish yeah. You are the boss.

F

F

Round number: six: seven! So you understand it's only you guys on the top, then what about the other contributors? We just started out. So that was my other concern as well. Like you guys, we are the legends of the Legends I mean it would take us not even years.

A

This Spotlight each.

F

And every day, you're opening pool requests and yeah.

B

That that's called spamming like a pro, um but but countries like but you're talking about total right in.

D

B

It's huge but shouldn't we just enable uh like um basic, like of not sure if we should call it filtering but enable people to see okay, who was in 2026 when he said like to check basically uh uh how to.

C

B

Specify your date, hey, I, want to see uh who was top contributor in 2019. So just uh you know what I mean so we so we basically generate a lot of different tables tables, so we have table per maybe even week uh per month per year, then we merge them into also a total all time.

B

So we have like um so like the the main contributor view is all time right um because like I mean why not even though I'm on the top, which I'm happy about, but it's like um yeah but we're showing all time. But if somebody wants to see 2018, that's a completely fine I mean uh I mean um like. Is there uh some limit? We can't do it or like, or you just don't like me and don't want to see me.

F

Okay Lucas, so, if I understand correctly right, so we can have like data from 2016 um as per month like okay from November. What's the top contributor December January blah blah blah until present, then we have a table like for uh overall a year like top contributor from 2016 per month, then top contributor for 2016 a year. Then is this how you explain it right, yeah.

B

Yeah like so, we have several different, uh like granularity of the information is on different levels like week month year, uh oh and and then total and then in the end, it's Excel, so we can put it into CSV and then put it into um like read it in the website and do whatever we want with the data merge. It show it different graphs, whatever views but like basically um have different uh different data, like not sure how else I can so.

A

C

A

um So like, according to what you you like your idea, um so the question is like, if I'm trying to like get you right. That is a lot of query to run.

F

ah I can reuse the script, but there are limitations. I cannot get data from um day or weekly, especially if it's data for the all data, the only data that I can get is like month and year. Those are the two tables that I can get like for previous years. Then, would.

E

It not be possible to store it all, just in a simple static, Json file. Do we even need like to run the script again? Because if you first like can't, you filter it afterwards, like processing.

A

You mean the past data.

F

E

F

E

A current because, when your fetch the current uh the current data, you can still post process, filter out specific users or specific dates or Etc shouldn't the website be able to do that on its own.

F

A

Yeah I think it is, it is based. It is based on the uh sorry, employee, I think it is based on the kind of data we get back. It does like includes um pretty much it's like days months and stuff yeah yeah so is.

F

A

Possible to live.

F

Okay, like I, explained the only three queries that you can make um is only day a period of this day until this day, then a period, the monthly then yearly. Those are the three chapters.

E

Yeah, so so during the build phase of modeling our website. So when you build the website, it wouldn't it be possible to for the build script, to pull down all the they said if it hasn't already installed in a Json file, a list basically locally and then once the UI needs to access it and filter it or whatever. You don't need to make any queries to the GitHub API. You actually have it.

E

It's only relevant when you're doing for the current inbound year right as far as I understand or something when you haven't fixed it before.

F

Okay, so one thing to be really honest: the data is large. Github data has a bunch like, for example, this data. This weekly data is like um 138 gigs, already to query this data that you are seeing your idea, so I cannot fetch like um the past data. The only thing that it does is to fetch, like in tables, so I have to create like one bit query, because the other ones during the whole data.

F

So it's easier for me like to add the dates or add the months that I want async API then export this data to an Excel file or a Json, but also there is a limit to this. It only creates uh I, think it's one gig Ace I'm, not sure I explained this. Also to you there's a certain amount of data that you can get Excel um from if you're exporting data.

F

It only gives you one gig of data because also I'm using a free version, I'm not using a paid version, and even if we were to run like a daily okay like give me from 2016 on this particular day and then store it into tables. We need also I have also limits. I can only use a certain amount of gigs of querying data on bigquery and after that, I need to pay, if which is really a lot, so those are either delete.

F

Those are some of the limitations that we need to consider, especially if we're going with that approach. So this is why I'm was also saying that it's either we can query monthly data and store it into tables. At least monthly data doesn't consume a lot of bytes.

F

Then we can have yellow data over like from the whole 2016. These are the top so like from 2017 January. This was the top contributor February those tables I can do, but if we are now saying like every week or every day, that's another different story that we need to consider as well. I'm, not sure. If that explains it.

B

Yeah um I mean in the end we don't have to have everything that, um with the first iteration, um even having this monthly, that that's already a huge, uh a huge thing um and in case of all the data in the end, we just have to run query only once right. So for the data 2016-2022, we don't have to run it every time uh every every week again the same query, because uh we just want to run it manually once store it and reuse it.

B

um If we um for this year, we can also manually run it for past months, um so we would only really do the monthly or even daily, runs starting from some specific date.

B

So on the cost level, each then I guess it should not be a bigger problem right.

F

A

So I'm trying to like understand um what we're going to have in the UI right. So we have this list of data. We have this filter drop down, which says say: contributors 2016 contributors right, so we click on 2016 and we get data for 2016 right. So we don't have to like filter by month for 2016. We just get data for 2016., then we can only filter by month. If the year is 2023 or 50 years latest like if the year it goes to the currency. We are.

F

A

Then we can say Okay free time a month.

F

Yes, we can do that like, for instance, right instead of running this group, run it monthly like from the 1st of March until the end of March. Then we get okay, we are the top contributions or we can run it daily, because what I remember is if we're running eight by month, it's not a lot of day, there's not a lot of bikes that get consumed and there's a lot of data that gets a spot exported so yeah.

F

We can do that for like present data, there's not much, but if it is past data, that's a different story.

A

Okay, now I have a question which means: how do you turn on rendering this data right? For me, like you, of course, you're gonna like pass it to an Excel file, but now, okay, are you gonna? Do like room 2016 passage to an Excel file, so we just have like an Excel file for 2016. We have an Excel five for 2017. We have an exam file for 2018.

F

And welcome to data science, This Is, What DJ is all about this. Ace I will create tables for you. You will get data a full table for 2016 17 18 until 2022, then for 2022 you will get a monthly data for January. February March is.

A

F

For 2023, you now get like four you. No, since we are still.

A

F

Yes, this is what I'm saying we can't have a table for 2023 as yet, but you will get like for months. We.

A

Have yeah yeah, okay,.

F

We have for February we're for March, then, by the end of the year, we'll tally everything today together and you get your table for 2020., because that makes sense for you.

A

Yeah for me and.

B

And in case of uh costs to understand again uh it's the query that costs not the data that you expose that you um yes x, x, exports.

F

B

Yeah exactly yes,.

F

Is the query that costs, because, when you GitHub, is going through a bunch of repositories, a bunch of data from different organizations so yeah? It's the query that costs, but the data to export it? No, it doesn't cost much as long. It's not like a gigantic data set or Legacy.

B

Foreign like we, so we don't so we because I assume that your query is already doing the aggregation of the results, like counting already the issues per can. Can we, for example, don't do it and just have raw data like so basically uh in case let's say: Jonas have or actually Ace had these 10 issues weekly. So instead of table with uh A's 10, we get actually 10 rows with A's and date of the contribution and link to the comment, and then we on our own just have a code that Aggregates it.

B

You know what I mean like more raw data than um that you can operate on I.

C

F

Understand so like.

B

So now, when you show the stable, there was um Ace and issue starts wherever 10..

B

So, instead of a is 10, can we get an axle uh or whatever table, basically um a table where, instead of a is 10, we have 10 times A's mentioned and there's a there's another column saying it's issue start it's a link to the issue comment the date when it was created.

B

So from that data we just have our JS, whatever code, that aggregates 10 Aces into one ace and provides number 10.. You know what I mean. So we don't get this aggregated data, but we get actually more granular data exported.

B

So we can access them because, like later for the UI I bet, people will say: okay, you say you have 50 issues. Yeah I want to check what actually 50 issues you have um and we will say sorry we just have the number we don't have links. You know what I mean.

A

But isn't isn't that going to just extend uh the size of the file like like I said there is a limit.

B

um I don't know.

F

And also the way your mesh like get a link to the particular issue that, as has opened or as is contributed right, is that what you're saying.

C

F

I'm not sure how to do that, because please remind like I explained to Ace. There's no there's.

A

No documentation.

F

I'm using I, don't know which documentation there's no documentation. How to do this in that GitHub is the rest. I had to figure it out on my own honestly, the only thing that I remember was like you can only query. By day by month, Empire and I had to figure out the rest of my own, so in links the only links that I remember that I can get is like um the links to the profile to the actor profile, not to the issue.

F

Then I will need to like find out if it is possible like to get the.

B

Particular issue that we are counting so I I suggest um I'll make sure that Linux Foundation also invites you to the dashboards that we have access to, um because they're doing uh something similar that you're doing I bet and I I'm, pretty 100 sure that they're using GitHub archive, because what else would they use so um you'll, be able to see what they have access to and I'll share with you the their Forum, because they have pretty cool people working there are super responsive, so maybe they will be able to help you like uh to guide you like, what's the best way like how they solve, because they were solving it for all the Linux Foundation projects uh doing different dashboards Etc.

B

um So maybe they have some answers. Also, there's this chaos project from Linux Foundation. That is like fully focused on like on metrics of Open Source contributions. So maybe they also have someone who can help out to to to somehow get access to it or maybe even GitHub. Maybe even Linux Foundation can give us access to their raw data as well, um but yeah there's there's a lot of maybe so um I I'll contact you after the call and make sure that you have all the accesses and access also to people that can help.

F

Okay, all right.

D

A

Yeah I think that's it's at least. If we have more data to like work with, then we should be able to like provide more information instead of just like putting the whole stuff. We have out there so mixes I, really hope it's you get like get it figured out. So um should I wait till you, you do that or it should be something we can like had as like um later on, like if you can get this current one like like. We all discuss like based on here.

C

A

Can you hear me.

B

um Like you mean we want to, you want to just go ahead with implementation, or what do you mean.

A

No like with this, no um because now, I have to like start working on the on the first iteration right. So I was thinking. If we, if I, can like just like, um implement the current rounds, we decided, um which is like having the data for um yeah like 2016, then 2017 till 2022, then half data for 2023 so current.

A

um If she can like Implement dots, then I could because I was also like I want. What I want to do is work on the first iteration and be able to like other contributors, contributes to it, because if I just implement the old stuff that they're not gonna, have a chance like improve themselves.

B

Yeah but the problems you can't really work on even first iteration, because we don't like uh whatever we have it's like so truly- is using some free, uh big um uh Jesus like I. Don't remember the name that stores the data and can we can query through it, uh which is yeah, bigquery yeah, so we're using some free versions. So we even we even here we have to answer a question like long term how we want to proceed because we uh it's gonna, bring some costs we have to like so like like.

B

Basically, there are so many questions even related to this, like. How do we do it long term? How do we are assure how we are sure we can actually cover that costs?

B

That's blocking you anyway, because whatever you're gonna work on you're, gonna, publish and then what like, if we say, we can't cover costs, because it's way too expensive.

A

B

Down the page, so yeah I mean I, see you're super excited about first implementation, but yeah. It's uh don't like sorry. You have to wait a bit anyway.

F

Yeah, it is true because, like.

A

F

Long term, we need to also figure out like querying. It need to figure out storage as well, because we cannot be exporting to um to excel all the time. So we need to also figure out storage as well like a viable place where we can store like, for instance, in the query Cloud. Then you can create data directly from bigquery cloud until to the website. So that's another thing to also figure out as well. So it's not a walk in the park as yet, and.

A

I mean that who.

F

Wants to manually do it like okay export, like when.

D

F

Also think long term.

B

Every morning,.

F

I don't mind: I will just give you the scripts and say: okay guys do with it. Bye.

A

Yeah, and, and also like the the old concern like yourself like, have a discussion around because USB brought us something interesting. Oh sorry, it was running up this meeting now, um but before Roundup you brought up uh something concerning like assigning, Badges and stuff right, and so we need to also like discuss which contributor get which badges. Like? What criteria are we using to measure what what parties someone gets.

B

I'm, sorry to bother you but like I, have to uh I have to escape the meeting.

A

Yeah, so many dimensions anyways, um so I will continue to discussing that in the next bit and anyways. Yes, so uh I, you don't still imagine I thought we're gonna have a short meeting that always happens.

A

Okay, uh okay, thanks for doing this call um catch you guys later in the next coming to meet. Him have fun bye.

F

All right, bye.