AsyncAPI Community, 23 Jun 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: AsyncAPI SIG meeting 26 (June 23, 2020)

Description

This is the recording for the AsyncAPI Special Interest Group (SIG) meeting #26.

Attendees:
- Colin Sullivan
- Fran Mendez
- Łukasz Gornicki
- Marc DiPasquale
- Mohammed Rizwan

Moderation:
- Łukasz Gornicki

Agenda & Notes:
https://github.com/asyncapi/asyncapi/issues/394

Chat Dump:
00:45:25 Colin Sullivan: Hi all, I need to drop off. Take care, see you next meeting.
00:47:07 Fran Méndez: https://github.com/asyncapi/tck/issues/16
00:47:20 Fran Méndez: https://github.com/asyncapi/tck/issues/15

A

Hey again on the asking API public meeting, so we have here me as a host and a I'm gonna run the meeting from not sure if you'll make it for the whole meeting, but it's anyway I did not expect that friend will be able to join. So it's good to see you from anyway, and we have Colleen mark.

B

A

Not going to explain your view, you should expect yourself.

C

D

C

Was a couple weeks ago right that your life is just a sink APR right. So what do you need to do? A harbor I.

A

Anyway, would not call it this week that he's now doing a holiday anyway, because it's just not much interesting.

E

But I'm gonna have to drop off in 30 minutes, but I was attend as much as the meeting is again yeah.

A

Sure thank you, um and we also had Muhammad that also joined in the past, so hits we have Colleen mark and Mohammed. So thanks guys a lot for joining so before I'm gonna jump into our agenda.

A

Like you callin, you said that you don't have time for the whole meeting, so any of you have something that would like to talk about before I jump into the usual agenda that you would like to talk about it now in case. The other topics are thinking too long.

A

You No, okay, I took it as know at least. Silence is always good. So let me then share the screen.

A

A

Can you confirm that you can see it perfecto, so there are just two items today. The first one is the give you a heads up on the status of the generator one zero release and also the JavaScript parser. Once you were release so the the generator we already breached release candidate number four and it did not get any update so far, so it looks like it's. It's stable, there's still this one issue here, but there were no objections for us to drop it.

A

The investigation I did here so I will. Just simply. We know that will not add this types generation before one zero, so I will just remove it from this milestone.

A

So, as you can see, basically the generator could be released already, but- and we already checked it with the with the github action with the officing api, with the with hab I think API that IO all works well. But now what I'm working on is making sure that the, because we also want to make JavaScript parser stable, so I'm now fully focused on parser to make sure that the to complete all the tasks that could possibly cause some breaking changes.

A

um So you know we don't end up with to zero too quickly and generator is dependent on JavaScript parser a lot so there's a milestone for JavaScript parser.

A

So most of the actually all the issues are started already, so the this one is already there's already PR waiting for review in this. This one depends on on the previous one, because I'd say there are some manual validations and that were not in the parser in the past. So before we can jump into task, then we first need to have merged 24 merged to make sure that we did I, do it properly and I can continued with it.

A

For the other validations, the mime types the PR is already opened, I create I, worked on it today and that's already. This task is a proof that it's good not to release generating one zero. Yet because the at this one 47 it's a it's gonna introduce a breaking change in depart into parser, because we change away the the custom parsers for open API or for offer of or for whatever they the way of registering those parsers changes and the last two. There are some open questions to this one and the recursive recursive messages.

A

That's another topic on the agenda, but yeah. That's also. We made progress in discussion, so we're pretty pretty active here. I would say maybe two weeks and- and we should be done and the flow will be that first we'll release parser one zero and then generate the run zero.

A

Any other questions too. So.

A

Yeah, but basically to summarize that's the that's the main focus now so most of the things most of the other things I'm, simply ignoring to make sure that we complete them some and yes, circular refs I've put it on the agenda because that's a I think it's interesting discussion and I'm, not so opinionated to have strong opinion in this topic. But basically, what or is it about?

A

Is that now, at the moment, when you drop in a a a spec that has some schemas with circular references, then there's an error because under say the maximum call stack size exceeded. Message goes in because we resolve the references, so the so, for example, the function that we go through all the schemas recursively to assign the the anonymous IDs for for schemas. It just never stops working, and then we have a error in the parser that of course, then enters up in the in the generator. So there are, there are some investigations.

A

I did in the in the dependency that we're using for parsing references and to my knowledge about these dependencies. There's no nothing that we can like no option that we can just take and use, and we simply need to discuss and and and agree how we should resolve the or circle circular dependencies. If we should just simply ignore them. But then it means that the refs will stay and we need to somehow come up with solution.

A

How we will how we will handle them in the parser, how they will end up later on in the generator interesting API document, because otherwise the the templates will fail because they don't know what to do with the ref fields and the other option is and I think I saw it in swagger or some other tool that people basically make a make sure that whenever the circular dependency is is discovered, it's marked in there in a in a string that a particular field.

A

It's um it's a physical dependency, a circular, a reference and that's it, but the but the rest of the object is resolved and it's returned as a si. Si object in the schema which works well and will work well in the templates, but that's a custom solution that would have to do and I'm not sure if it's, if it's good, because for documentation, it's it's perfect, because that's what I saw and I think it's it's the best approach to to just reference as much as we can and then those silk or our references circular.

A

The fields that have circular reference just mark them with a string description that it says you are but I'm, not sure how it's with the code generation. So that's pretty. That's pretty open question for me like, if I'm generating code for any application. What would I expect in those two how to treat those say for all references?

A

Any and ER I'm super open.

F

Which has structures for recursive, you know for ways to find recursive structures right, so we should define there. You just probably need to do what you said. It's like. We do read only the first time market as circular somehow I don't know how yet, but next time your parser is gonna, be reference. The Doha Rev right.

F

If the text and it's being already its market is simpler, anything stop there. Okay right so yeah.

D

We still have a way of stopping.

F

Somewhere right, some.

B

People say that I.

F

I've had this conversation before with open API and because yeah it's the same problem so and some people say that okay.

B

So the reference.

F

Absolute form levels, you know I'm gonna reference, the resultant references formalistic only so you never entered them in the soup, but yeah. Why four times and not fly for not three. You know it's like. Is it like a random bender, so so yeah I will probably go for more reasonable or.

F

Yeah or reasonable solution like how can we do it this way? Because this and that right and that's what people or tooling creators can't expect, and they will have to do it if you, if you receive a $1, a day, filled with dollar wrath, plus the sequel through plus about other information. It means that if you want to do something there, you will have to do it yourself right so so yeah.

F

We will not provide any any value there other than just market, because we don't know what you're gonna do with the circular reference right for the clock you say like for for documentation might be. You might just mark it as circular and just put the link to do the finishing of the structure somewhere or yeah.

B

F

Marketing you I, but for cogeneration in Java you'll did one way it Duchess way or some type script. In this case you do it another way so yeah and go in different ways.

F

So there's we cannot help other than just informing the tooling creator. That is reference to Sigma, and here you have the value resolved or yeah more than the value is often said. He really have where to look at in the current document. Where can you find these? Well already? Like an example, you have there in the or that.

F

Sorry forget its name, so that that is one on the other example gear that there's a.

F

Head first, for instance, if the first book.

A

Which you mean example from the description or yeah.

F

For the description exactly so, this one produces the first. The first one is components, chemistry cursive, but we we are really ending recursive in recursive right, so so yeah. We shall have a way like just say where you can find this. These structure, which is a living document.

F

In the case of the commented in the commit to use case, it's same thing, it's it's one level deep, because it's a property but Athenians you shall have the recursive, take it, take the structure somewhere in the document or somewhere, if not to document, at least somewhere in there in the list of schemas, that that the parser is going to give you back. Why? Why am I I'm trying to keep saying that it's gonna be already defined on the document?

F

It's because this example has components schemas, and then you see it with person on the definition right on example, but this definitions by beating a level five right and then the recursive definition by being a third different file. So you can you?

F

Can it assume that it's gonna be on the recent get fired on the initial as you keep a file and but the information that the person gives you as a truly creator right, the information the parcel give you gives you is like, for instance, we have a, we have a a function, I think it's called all schemas some plan, this right in business of schemas. We have all the schemas identified by nav. Whatever is so. Maybe when it's resolved at Dollar Rev, we can put some information there.

F

We can add some information there like take in the list of schemas that you're gonna receive when resolving this. You can find these these recurse recursive, which is it's just the object, color recursive in this case right. You can find it at least behind this ID when ready it's generated or the recorded.

B

F

I need so I, don't know. If I'm explaining myself you.

A

Know, that's fine! It's it's just the matter of making sure also that the information that we provide it's easy to parse, right and yeah.

F

In that template.

A

You can interpret it yeah.

F

Let's not Francis, let's not assume that the reference is gonna, be on the same document. So let's not use Jason pointers like Jason pointers, starting with that a stack with a pound symbol right because the reference might not be there right and and also there, the recursive.

F

The recursive structure might happen involving three four five different files and that's very possible. The moment.

B

You start sharing common.

F

Objects across your company like a way to find something that then this object is a way to define this thing plus something else. Then these are the object. Is these two things combined plus a field, and you start playing with this it's easy to to make recursive structures involving three for more fire threats, so so yeah so anyway.

F

The other thing is that whatever information we put on the on the in this case on the JavaScript object that we return to the object to the user, it should be a problem, a custom way that we use to make reference to today to this recursive object and not I'm, not assuming that it's gonna be defined there on the other component, schemas right.

F

A

This makes sense, yeah I think it's pretty similar. Already to what so you can. Can you see it? Yes, so that's what I'm getting when I was doing. The research I was I, so basically in JavaScript, json stringify does not support circular references and there's a library that does it and what this library is doing, its the the field that introduced circular references, that's what it gets so what value it gets.

A

So it's pretty easy to parse I would have to check how it deals with the 50, as you said, with those typical serial cooler references across files.

A

But that's you can see here that should be already super useful and easy to parse right. So you get a info that it's circular and also in Oslo mission or so.

F

Also be aware of the different house that can be in super references right, because in this case, what we're doing is the recursive object is defining a field that has a reference to papers right.

B

It's a single off, let's.

F

Say but you might have multiple blocks, let's say it's like it's more complete, complex structures like one of its making a reference to another one, and then it's making a reference to another one and then from here it closes the loop. So you still think we're and again- and these are the ones that are more the most difficult to detect, usually because this ones are easy, you spotted at first sight, but the ones having moved into hot. It's usually the past and across Finance, of course, are the most difficult to spot.

F

This I would think it is. It took us, go ahead, probably add some tests say yeah.

A

We will have to have really good. That's you're here.

F

Because then, we.

A

Okay, anyone wants that something.

A

F

B

Not looking at you mark.

F

Cool same here is it because we are working with a very dynamic language like it's yes, fully dynamic.

F

So the cool thing here is that we are learning a lot about the problems that you can find while parsing this kind of documents so yeah these learnings will be applied to general, a parser.

F

Parsing the features, but it so we were gonna, have the same problem. Basically, that's where you think.

A

That we're gonna fold it only one ship.

F

It doesn't have to be, it doesn't have anything to do with types in this case.

A

Okay, so that's it if it comes to the agenda so.

A

Anyone has some questions to pigs, handle.

G

G

Last meeting we had, we talked about Java passes right, that I told that we we can contribute and we can create a Java passes for that I think Fran told he will talk with some other guys to check whether they already started working on Java, masa, yeah.

A

From you were supposed to talk of yourself, so I think yes.

F

F

A

Was also down in the evening so I think they are now working as Bank whenever they changing something need have, they have to switch it off.

F

A

You're gonna share something, should I already stopped sharing and give you a host.

F

These two issues.

F

We're actually the result of our conversation two weeks ago, I think and.

F

So they'll go they're gonna work at an open.

A

Source of this and.

F

They are now working on making the parser the jungle parser they created 100% compliant. That's why you see one of the one of the uses about test cases and and I think once they reach this level, they will open sense it, but honestly.

F

That's all I'm Kevin, that's what I can say like it's, it's just it's all in their hands. It would be great to have probably innocents like a miniature someone else from the team to join one of these goals one day so sorry.

C

A friend can you explain what what it is like? Maybe it's a minute explanation with yes,.

F

C

F

This project is Reapers about is a TCK is test compatibility, kit right and the recent point I'm pointing there there it's, because this is a project bill. That's contributed right and it's a way to test the parcels right test, the compliance of the person. So we are we're now so far, testing our parser and their pattern in this in this project, and you can see some some details.

B

That percentages.

F

The compliance and all the stuff right- and the thing here is that they they haven't, opened the parcel. Yet as far as I know so.

D

F

It will be great to have them someday and explain how to use it or when they are, they planning to release it and.

B

Opelousas it so.

F

Yeah yeah yeah because that's a that's gonna work on Bank Street in Java as far as I insert up Java Scala, but yeah the jvm language will be competitive. So so yeah. That's that's why I know that's that's what about this I information I had from there is they I.

F

F

Actually, it will be great that they open source it even before it's perfect, let's say ten percent compliant, so we can all help.

A

Question so doesn't mean that we'll also have to kind of agree on a common error types yeah.

F

Way to communicate errors, yeah that'll be great, so probably.

D

Yeah, that will be.

F

A breaking change for parts of it, so this is gonna probably take time and it's going.

A

To take time for showing up okay.

F

Probably just this number first version two, we will include this way of communicating mirrors which, unless you are working on a tool that sparse in basic API for another, something like solar system, for instance, unless you're you're working on something like this, you should not be affected.

F

Actually, no I'm not gonna, be thinking about it. It's not even so, it would be effective because you are exporting to the places API not important from basic API yeah. So.

A

You should not be.

F

B

F

A

Because in the generator is just gonna be bumped to the new version, but it's not gonna be a breaking change for generator. It's gonna be all that breaking change for the parser yeah.

C

A

F

Yeah I like to be losing often right, the fact that we're launching this a week that we can do movies person to you know six months or.

B

F

The thing is that we will have to maintain. Also person won't, like you know, security, patches or.

F

Support and the support formation one, but we will have to be meditating two versions, but yeah I, like this style, for instance, that formed for this sharing some long-term support.

F

And the support of old versions, right I like what know Jesus to in phrases that every today they main thing they provide modern support for one Russian. The next version is not going to be long-term supported life, it's the ether, commercial, the eternal verses, even dampers on long-term supporter and then the off or not, but not meaning that they will be supported from only six month or one year, but that they are not.

F

B

F

For sure and and the answer that you can keep iterating without the fear of breaking something and then what people? What not GS says.

B

F

Every adopter production, the doctors is, it useful, adopt an evil version for port reduction, never enough version unless you're playing with it. But if you want long term support for production you to adopt 10, 12 14 right, not 13 or so so.

D

G

So I understood that Java parser is doing by the team mules of team right, but it is not it open sourced exactly so. They.

F

Were working on this, they were going to open source it I'm, pretty sure I cannot I cannot guarantee because.

F

I'm pretty sure, yeah and then finally, this Muse of this company.

F

G

It would be better to wait for them to open source it instead of creating a new project as Java right.

A

D

At the end, how.

A

Much time you have to wait because looking on those two issues, it can take them still few several weeks even before.

F

A

But you can create a like middle solution that will serve your purpose and then.

F

How about that and see if they can open source pika before this is finished, we can probably help them.

G

Is there any timeline or I mean any dig that they will open source it? Because, if most of the AP gateways, basically they develop their project using Java?

G

So if they want to use the async API, so they they must have this Java passer, yeah.

F

G

So, based on this information, I mean the information, then it will be available as an open source. Probably typically providers can decide and you can rearrange the pipeline right to give the support for a sink. A PA.

F

That's all we can do so far from outside, also something that we were planning that we're not working on this yet is working on our general parts of ourselves.

F

Solution but yeah, depending on how good it is the muse of word and and how often this is fixes maintained in all these things. We probably interested advise people to product their parser if its auspices, of course, that well yeah.

C

I mean friend, I guess what a suggestion there potentially is to let whoever know soft don't know if it's Paul or the contact is that it was about this, and let them know that you know there's there was interest on the call and that potentially it sounds like I'm sorry is Mohammed is your first name right: yeah, okay, yes,.

G

That the Mohammed and his.

C

Company might be willing to like help contribute once it's open source, so maybe give them some incentive to open source it earlier right. Yeah.

F

Remind me you work for software AG right. Yes, yes, fine, okay, okay! So that's that's the one change just think as after the call and let's see if we can arrange this and if we can find a solution, because I think this is gonna, you can figure some slack. Maybe if you're there are you arguing our slack workspace? Oh really, yeah, yeah I'm I mean it's like yeah, okay. So so, let's let's do this and then we can find a solution.

F

You know if it's you know adopting or convincing music to open source it sooner or or creating around around Java parses right. That will make whatever works.

G

Okay, okay friend and I will probably ping do.

F

The things that we cannot do much more.

A

C

Questions topics no I'm good I, figured I'd drop in and see what was going on and I'm. Glad I did to hear that v1 is coming out soon of the generator enforcer, so I'm good, yeah, I.

A

Own, my own cannot wait until it's done. I have to relax run away from code for a bit. Just need to finish this and find me to some of the commutation for generator.

A

I'm spitting, like yesterday, I was coding until 1 a.m. like the whole day like crazy, like I, have to finish this one that were another one and maybe one week, maybe two weeks and it's done I have to rush.

F

A

That's how it case.

A

So before we finish kind of request to you guys, if you're already on our slack so please go to generate, generate general channel and there's a information for me about a new blog post. We have from James Assange API blog. So if you can help up help out and both up those to the blog on Reddit and hacker news, that would be.

A

Danke Schoen wiedersehen, thanks for joining remember in two weeks we see each other again in the epic frenzy.

C

Okay, so two weeks from now is a pack and then two weeks later is this so.

A

It's now always gonna be 8 a.m. UTC or 4:00 p.m. you.

A

F

Start doing weekly these weekly meetings or this meeting weekly right because otherwise people from the from each of the areas, let's say from each of the sites like America and so right. They will have to wait for one month until the next weekend. Right.

A

Sorry to say our faces like waiting for one month to see our faces. That's your.

F

Center, fine, but because you can join both, probably but yeah Australia and America. That's gonna be hard for both join all of them. So yeah.

A

um So, like officially now closing thank you or seonsuk.

C

Enjoy your vacation friend.