AsyncAPI Community, 22 Mar 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: AsyncAPI SIG meeting 69 (March 22, 2022)

Description

This is the recording for the AsyncAPI Special Interest Group (SIG) meeting.

https://github.com/asyncapi/community/issues/294

Chat:
00:05:28 Jonas Lagoni: https://github.com/asyncapi/community/issues/294
00:07:35 Sergio Moya: https://github.com/asyncapi/spec/issues/735
00:11:58 Jonas Lagoni: https://github.com/asyncapi/bindings/issues/113
00:14:18 Jonas Lagoni: https://github.com/asyncapi/spec/issues/590
00:22:52 Khudadad Nomani: I want to take this issue. but I am not sure where to start.
00:27:20 Jonas Lagoni: https://github.com/asyncapi/bindings/blob/77f6f7744d79619e1e663bda65bf7a535c4c21fa/kafka/json_schemas/message.json#L14
00:28:38 Sergio Moya: JSON Schemas for the Spec: https://github.com/asyncapi/spec-json-schemas
00:31:10 Sergio Moya: Need to leave for 5 mins
00:31:11 Sergio Moya: Back soon
00:32:14 Sergio Moya: Back
00:34:36 Jonas Lagoni: https://github.com/jonaslagoni/spec-json-schemas/blob/567ba22ad754591c7191dfdf7a22819317914500/definitions/2.3.0/message.json#L52
00:36:51 Jonas Lagoni: https://github.com/asyncapi/spec/issues/590
00:38:58 Aayush Sahu: Take Care Lukasz :)
00:39:12 AKSHAT NEMA: is the situation getting worse in your area?related covid
00:39:48 Vivek: Take Care Sir 🙂
00:41:07 AKSHAT NEMA: Anything related to mentorship program?
00:41:10 AKSHAT NEMA: updates?
00:42:53 AKSHAT NEMA: Anything you want to discuss related to the issue "Docs Feedback Form"? :) We had a discussion there

A

A

So hello, everyone, I'm sergio this- is the community seek meeting on async api? Remember that this is being recorded and.

A

Yeah uh today there is no lucas around and I'm trying to do this so yeah bear with me there with me. um Okay, if you want to introduce yourself, if it's your first meeting, I don't think there is any new person here, but if you want to take the opportunity feel free to introduce yourself right now.

A

Okay, also, if you want to make questions, upfront or yeah, feel free to interrupt them, make them or otherwise wait for the end of the meeting. So you can be also making them.

A

Okay, so what's the agenda, let me share the screen.

B

A

Okay, so this guy here sergey moya, which is me I'm going to talk a bit about the new release that is going to happen next month. This the 2.4 release and then jonas is going to talk about bindings and the relation between buildings and schemas.

A

um So, let's start I'm gonna start speaking about this um yeah release. 2.4 of the spec is gonna happen next month.

A

If you don't know about the release currents, this is what we plan to have. We had one release in january. We are gonna, have another release on april, another in june and another one in september.

A

uh This doesn't mean it should happen, the very first week of each month or the last week right. So it's just around the month. It depends on the the work we have to hey lucas, hello.

A

uh Yes, sir fox emergency, it's fine do you want to take it or I can continue it. It's it's my turn now so.

C

Continue to continue.

A

So, as I was saying, uh 2.4 is gonna be released uh next month. That means april. uh There is already this um sorry because at the same time, I'm this issue created um with the whole process of releasing.

A

um There is also a guideline that recently updated dail, which is amazing because everything I had to do it's here, so I just had to follow the guide.

A

So there is also a new issue template that looks like this, so even the template of the of the issue I didn't create, that is just automatic, um but I had to do something right, so we already have the release branches, which is the 2 0 to 2 0 4 release.

A

There is one on each main repository involved in the the release, which is the spec spec json schemas and the parser gs.

A

So what that means is that anyone can start contributing to those release.

A

Any new feature that comes and that it's a candidate for being released in the 2.4 can directly target. The request can directly target this branch.

A

I also created the draft notes for this. uh It's not linked here yet, but there is nothing basically just the announcement and that's it um what else? Well, here it's everything I did. So all the branches are up to date with master as well.

A

And the most important part here this is something I'm adding on top of what the guidelines say. So I wanted to put all the work all the potential work that can be included.

A

In this version here, um the only one confirmed is the very first one. They make servers variables object available under components. This is already all the pr's are. There are reviewed, so it's just a matter of merging them.

A

The others are under investigation if we can or not uh make it for 2.4 and yeah. Basically, that's all I'm not gonna extend this a lot. There is nothing else to say I guess, uh apart from calling maintainers to help with the release, once everything is merged,.

A

And also contributors to take this opportunity and make their contributions, so they can make it for 2.4.

A

C

So basically, we know that 1 2.4 will happen because there's one feature at least.

A

Yes, yes, this one here.

A

And potentially I mean or allowing grouping, servers or the introduce of environments it could go in, they both are very similar, so they can be merged together if it makes sense- and there is no a lot of work there to do- but we'll see.

A

Cool, so if there are no more questions.

A

I guess jonas is your turn.

D

Yeah, okay, so let me find the issues.

D

I guess we can start with this one, so there is a proposal for moving all our json schema document for our bindings.

D

To our main json schema repository where the json schema document for the spec itself lies.

D

There has been some discussion back and forth whether we should do it.

D

And I think the consensus is more or less that we are removing it, but there is still some things that needs to be figured out.

D

One of magic's concerns is that once we move the bindings is to how we integrate those json schema documents into the main json schema spec, and whether we should do that because, currently, with our parser implementation,.

D

We kind of do the validation in kind of like two or three steps where we validate the entire spec and then not even sure we actually validate the bindings at the moment.

D

Now that I think about it, we don't we don't know.

C

Because we don't have access to the json schemas right.

D

um But the idea is whether we need to integrate it or whether we should.

D

Do this two-step kind of validation, where we do it separately?

D

The next question that comes up in terms of whether we should move it is well. Where should they be moved to and how.

D

How do we correlate that to the main json schema spec documents, uh and this is a discussion we have in?

D

um Let's see if I can find it that we have here, because we have no idea yet how to actually do.

D

Like how we tie our bindings to a specific version of the spec.

D

I don't think anyone is championing this at the moment, but there is some suggestions to do it in terms of compatibility between versions. So, for example, for version one of the bindings. We say that we are com compatible between version two and let's say 2.4, but anything beyond those the bindings no longer are compatible.

D

uh That is leads one suggestion at the moment.

D

um So if you have any thoughts or ideas, please please bring them to the table, but that that is more or less the current state.

C

So basically, the versioning uh lack of the answer for the versioning is blocking us from any decision, basically right, because even if we want to move it to json, add the schemas to the to the other schemas. We still don't know how to integrate them with the main schema right, like um yeah.

D

One thing we could do is not care in the beginning same as we have with like how it is now basically and just place all of the bindings in a bindings folder, and that is the same as like the current setup that we have in the bindings repository.

D

So in that way it we don't make any breaking changes and we don't make any decisions now. But it's something for later.

C

D

Doesn't give us.

C

Anything right.

C

I mean we just changed the location of the files, but we I mean what you're saying is that move, but not integrate with main schema or.

D

That's at least the beginning, yeah like how to get them moved afterwards. Of course, we have to figure out how to achieve this.

A

Yeah, the point is, I guess, you're saying this, because we can work in parallel in two things. Right like moving is one thing and the personing is another thing right.

B

A

We are gonna, fix or fix or introduce the person on bindings, which I don't know.

A

Yeah from from my point of view, versioning is the critical part here, as lucas said, so if we don't fix that the effort of moving can be postponed in my opinion and instead working on the person unless someone wants to take it, you know why not.

C

Yeah, I think that the biggest like, at least for me in case of bindings, whenever I take some uh voice in the discussion, is the the problem. Is I I personally cannot. uh I don't have a vision in my head like how it would look like now like we, I know we. We had this discussion about versioning, that you have to mark that the binding is compatible with spec between 2 0 and 2, 4 or whatever, but we are basing these opinions on.

C

Like um not on a real uh example, I mean, so what I'm trying to say is that we don't have a nice example like a simulation um that, for example, okay. I have now a spec, let's say with kafka bindings. I have a um nice api spec with kafka binding um and let's say we, we enable validation, okay and then we don't have any simulation like um like some fake simulation on how it evolves over versions like let's like we. um How do we explain it?

C

Like you remember when you were working on this um intent-driven api, you were kind of simulating like um trying to make some breaking changes in the spec to see how it affects the api, and I have a feeling it's something like this. It's also needed in the discussion about versioning of the bindings, because uh we can't really.

C

We can't really say how this versioning should look work. Even this idea with saying that binding is compatible with version two and two four, I mean it's just an idea, but it's not based on any any any example. Actually.

C

So yeah it would be nice to have someone who can jump in then.

C

A

It is a really interesting issue to work on. To be honest, I'm not saying I'm gonna be the champion because I have no more part with, but.

A

But I'm encouraging others to jump in.

C

But this is not related to 3-0 right, it's um it! Why did it like just uh just curious? Why it popped up just now like.

E

C

Because we don't have as you you noticed on us, we don't have itself yet so.

C

How did it pop up like, where did it came from.

D

It was because of the we work with splitting out the definitions I kind of wanted to have our json schema documents in a kind of stable state, and now we are kind of swaying, what's called like not really sure how to actually achieve what we want to achieve, and it was through those thoughts that I've kind of wanted to bring it up.

D

I think that's does anyone have any questions regarding those issues or.

D

I think that's more, that's my issue. I think.

A

Maybe it makes sense to to write in the issue. Sorry exactly what we have to do just to avoid people to have to read everything, because we typically don't know.

A

What are the features that it should fulfill right, like tooling, should be able to get one particular or compatible building version and well at the end. It's only only this one. I guess yeah.

D

I think I can not make a conclusion, but make a too long didn't, read, yeah and try and simulate how it actually works with when uh the spec changes and see if it makes sense.

D

But ah cooler that would be perfect yeah, so.

D

Yeah where's the start.

A

One thing uh it doesn't have to change if the spec changes right that doesn't have to do anything with uh with a new version of the spec. I guess necessarily, but it depends. I guess what I'm not sure what lucas mentioned about simulating. uh What's the relation, so I guess it's simulating how it will look like right, not trying to look uh at this spec on looking for breaking changes.

A

It's just how it would a proof.

C

Of concept, I guess right yeah, it's simulating! Basically, um so, let's we have this account. This super simple account example. I think right, um asking api document um that is by default, loaded on studio, I think, but yeah anyway.

C

I think you know what I mean so there's a super simple document and enrich it with some kafka, binding information like just just one, one specific uh information that is now in the kafka, binding and and it's state for and it's let's say it's asking api document, 2.3 and and some kafka details and now and now just simulate like um basically try to understand that um you create this document and all suddenly we release parser that supports binding validation um and so, for example, your document is, is working fine with the with this parser, because the the parser contains also the binding schema.

C

It's integrated with the main json schema. So basically the aj um aj file, uh another json validator ajp, agv, yeah agv. uh It consumes uh it's. It compiles the json as an api, json schema, but it also compiles the binding schemas because they're referenced um so basing on the information from json schema.

C

It validates your document assembly by a document, and it's it knows that, like um it's valid, because your information in kafka binding its um its string, whatever name of the property and then um and then like in version like, um let's say, um dale or lauren, um add something to the binding in the in the binding repo um to the kafka, binding, let's say rename the field and then once they do it, they make a change in json schema and the json schema changes in this.

C

um uh Basically the parser, because it compiles new schema and and first of all, like what happens with the document it starts failing, even though it's still 2.3.

C

So in theory, it's um it's valid document, but it's gonna fail because there's a there's, a problem with the name of the of the um kafka binding property because it changed so from that point um kind of simulate. What has to happen like what has to be changed? What has to be introduced in the schema so that your document is still valid? um It's still valid and still validating um how to say.

C

It still successfully uh validates, um therefore asking api 2.3, because um it's still valid document, it's just the binding version changed so um in the binding. Excuse me in binding. We also have this property where we specify the version of the binding.

C

So maybe that's enough actually, and we don't have to do anything and it's just some additional property on the json schema but yeah.

C

I without example, for example me I don't have a clear answer: how how that would work, and that's what I mean by simulating like um just answering these questions.

C

Because that would be the easiest like without um um because there are many different ways how we can integrate these bindings into the into the parser. uh It doesn't have to be through references in the async api um in the async api um json schema. It can be like. um um I think, machi suggests that it's just separate and it's just another step of the validation that bindings are treated like um something individual.

C

But that's that's that this is adding complication. I think the best is if it's um just treated as a part of the of the main async api schema.

C

And that's the simplest one and yeah. Basically so good answering your question like um so that's how it works like you have um in the spec. Json schema. As far as I remember the name of the repo we have fasting api schema, um it's included in the parser in the runtime parser agv library compiles the json schema and is able to validate.

C

As an api document, basing on the json schema as a first state, because there's there's this some additional manual validation uh logic in the in the parser, but the this, the core validation happens with hgv and and we need to answer like um how to how it will work. Once we integrate binding json schemas with asking api, json schema and start validating with agv.

C

How we make sure that myasync api document 2.3 with kafka, binding 0.5, will still work with with parser.

C

Even though we release next version of the kafka binding.

C

I hope that makes sense.

D

An interesting problem as well as, for example, the kefka bindings, actually references.

D

This schema definition in like the kafka binding itself. I send a link in chat. So how would that work when we have multiple versions of the spec?

D

But here it's basically only targeted a single specific version of the spec.

D

I don't think it needs to be solved in the same issue, but it's kind of like.

D

I actually didn't see this.

A

C

Yeah, but this one is um that's how it works now, but it can of course be um changed once we move bindings into the into the wrapper. We will not um point to some specific version, but the current one right. It's going to be a reference to the file system.

C

You know what I mean, so it's not like now, it's like 2.77 just to be sure that it's not gonna break anything, but once we have this, uh for example, this message dot, json schema is in the spec json schema repo.

C

Then it's gonna reference, a schema definition from from the repo, so it's gonna be released together right. So it's always gonna be the latest.

D

What about earlier versions? uh That's where the problem kind of seeps in right, because if the bindings like, um if the 0.2 version of the bindings are paired together with 2.0 of the spec, but the spec changes, major version so does the bindings, because whatever is not compatible anymore.

D

How do you reference the two point x? Basically version of the spec right in the old findings yeah. So we we're gonna have some some issues there, probably or issues yeah.

C

So yeah, that's also a good point that it's part of the investigation simulation.

D

Another thought that just came to mind is our. I don't think we can merge our splitting of json schema um files because in the current setup there is no way of the ah okay. Let's rephrase that when you define a payload for an operation or for a message um with the split functionality, it assumes that it's always the async api json schema variant.

D

But if you use a different schema format, it still validates the payload towards that.

D

I think, if I'm not mistaken, did that make sense, probably not.

C

Was it the question.

D

It was a question and.

D

I think it was a split between both.

D

D

ah Oh mind: it doesn't.

D

Yeah, that's me, rambling, forget I. I was fearing that we might be incorrectly validating some documents, but I don't think we are.

D

Hopefully yeah we are not because we never changed. The payload probably never changed this one. This was kind of what I feared were now being validated without checking the what's. It called the schema format.

D

But it is not the case so forget. I ask.

C

So the topic is covered for now right. Could I you more or less know, can you somehow? Let us know that yeah you know now you understand everything and you know how to start.

B

uh Okay, thank you very much for uh explaining. uh I will look into issues and then see. uh I will take note of the things that uh you have explained.

B

uh I will definitely see what we can do about that and then what needs to be done and then what is still not clear in this area or what to do about like, uh like the pointing of a schema to the json, uh pointing of uh a bonding schema to the expectation schema? How can we resolve that? So I will read more about that related issues and I will definitely comment on those issues or I will create a new issue to explain the situation or maybe get some point on those as well.

C

Okay, um jonas, did you link? Did you provide link also to this discussion in the spec repo about the binding versioning? I think we had some.

C

uh Discussion related discussion, not sure if it's mentioned in the issue, this.

D

C

Yeah yeah yeah yeah yeah from gerald yeah. Exactly so also, could I have a look on this one and yeah like, um as you said, like just ask questions um as you can see, these issues are pretty active.

B

Okay, thank you.

C

C

Do we call it a day?

C

Yes, I would definitely want to uh so. We covered two topics from the agenda. Anything else um from anyone.

C

No um sorry how you record it right.

A

C

Okay, cool and you normally could be a host when you started it from slack right.

A

C

The host okay cool- and you are cohort.

A

C

C

Which is a good news by the way um and yeah sort of works for yeah uh for not organizing it properly, but I had a quarantine until yesterday and um right before, meeting police came in to check me if I'm still on the quarantine. So I had to spend some time with them figuring out why I'm still on the list of infected.

C

Yeah, um that's it if it comes to interesting things that happen.

C

ah Thank you now, I'm good, I'm good police also left me. They didn't hit me uh for walking with my dog. So that's good.

C

ah No, no, it's not it's actually getting more relaxed, but it's um we had some hiccups in infections because of the amount of people um evacuating from ukraine.

C

um But yeah it's I think they are actually even removing the uh requirement for masks in few days so spring is coming talk of it and then again september october, we're back in in business.

A

Everyone wants holidays and coffee as well.

C

Yeah, of course, like coffee is tired as well.

C

uh Thanks for warm um uh words folks, um so yeah, um thank you sergio for jumping in and thanks for uh showing that I'm actually not important. Then you can do the meetings without me. It's a good signal for the future um and.

A

C

A

It was a pain, it was very difficult. I promise lucas, you have.

C

To be here next time, no, no! No! No! I mean only first uh time. It's it's difficult! Then it's gonna get easier, um so sergio you will share somehow uh recording with me. Of course, um I'm gonna upload it to youtube um so yeah before we finish. Five minutes left any final questions from anyone joining who anyone's anyone wants to clarify. Ask about something that was not in the agenda.

C

From my side, I can only add that, with the conference.

C

I have no idea: um last week's were not really easy to focus on long-term plans.

C

So yeah, um I plan to jump into the topic in the next days um in the in the mentorship program. um Yes, similar situation, um so we were waiting like two weeks to see how people react to it, but I think that this mentorship program has the most interest in the in the community and that I've seen so far. So um it's I just need to summarize um what people um said about the program.

C

um So I need I will just write a summary there, this or next week, um and we need to because it's again about spending money of of the collective.

C

So I I need to summarize it and give it up for the voting of the technical syrian committee so um to get an approval that we will do this program this year um and like what will be the rules but but yeah, you know more or less how awesome api works, so um um everything will be there publicly done nothing hidden. So I hope that we will have final answer before the next meeting in two weeks, maybe already with devoting any place.

C

But I guess like um wait um anything you want to discuss related to the issue. Dogs.

C

um I don't know, is there? uh Is there something we are we stuck there? Sorry, I didn't check it uh lightly. This feedback issue.

E

Last comment: we are actually waiting for the alandra's view on the discussion right.

C

Yeah, I don't remember so. Alejandra did not yet say anything right.

E

Yeah, I think you want to discuss.

C

Yeah, okay, so I'm gonna, I'm gonna, chase um alejandra and try to chase through different channels, but yeah I see it's. Basically, her answer is blocking the um further development, so yeah yeah. So first of all alejandra. If you're watching the recording, uh we need you but anyway I'll I'll try to talk to her on the separate channels.

E

Actually, I also want to discuss something that means discussed in the previous community meeting that yeah you have made uh regarding for the video discussion right for this particular year. So while I was on one of one of the one more issue regarding creation of the newsroom section, but I think we have also discussed there's something regarding the layout of that museum.

C

um So I mean I'm not aware of um what's happening with the newsroom on websites, so um I know.

E

That there was a plan to have.

C

It, but I have I I personally and I'm not watching the issue at all, I'm not even sure if there's apr already or you're still in the discussions in the issue.

E

Yeah, I think there was a couple of discussions between that time and I see.

E

It should have a particular section devoted for the youtube to have a particular content so that we have a live streams and all our recordings to be mentioned there and all our folks in the website should see. What is the present updates about the meetings and all the videos about.

E

Although I haven't created the pr yet.

C

um So yeah, I can only advise you like uh try to ping folks on the in slack and um because, like like, I said like I, I left it um so me personally, I I love the idea of newsroom uh to make it super easy to to access this like the live streams or even the news about from from twitter, etc.

C

But I I I left it for, uh for mache and and and messi to um to like basically lead how it should look like, um especially that design is not my passion, even though some people say it is uh so um yeah. If there's something like um blocking you there, just please being again in the issue, uh because sometimes these notifications just are are lost in uh in the float of other notifications and or just being folks in the channel. It's it's completely fine.

C

If you do it, um if you die ask directly in the channel for uh for feedback in the issue, um we just have a lot of these um issues across different repositories, um and it's completely fine, if you, if you ping people um in public and ask for for opinion, um I personally prefer this approach.

E

From your site,.

C

um Okay, so it's then at least in poland, um thanks uh to all you of you folks for joining um yeah, sorry again for uh having it not where it was planned, um and I hope today or tomorrow, and the the recording will show up on on youtube.

C

Thank you sergey again for liking, helping out.

A

C

And um see you in two weeks, folks.

D