AsyncAPI Community, 29 Sep 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: AsyncAPI SIG meeting 32 (September 29, 2020)

Description

This is the recording for the AsyncAPI Special Interest Group (SIG) meeting #32.

Attendees:

- Paul Taylor
- Fran Mendez
- Jonas Lagoni

- Ben Gamble

Moderation:

- Łukasz Gornicki

Agenda & Notes:https://github.com/asyncapi/asyncapi/issues/443

Chat:

00:16:16 Jonas Lagoni: https://github.com/asyncapi/dotnet-nats-template/pull/1

00:19:26 Fran Méndez: nats.io00:19:30 Fran Méndez: https://nats.io
00:22:42 Jonas Lagoni: https://github.com/asyncapi/ts-nats-template/pull/1

00:31:49 Fran Méndez: https://github.com/asyncapi/generator/issues/414

00:40:30 Fran Méndez: https://json-schema.org/understanding-json-schema/reference/numeric.html

A

Okay, recording started hello, I'm mukash, um it's a public asking api meeting for those watching recording. uh Today's meeting uh um is with me lucas, fran, paul who joined us from australia and jonas who joined us from denmark.

A

um We have, I mean we had two items on the agenda, but also paul a few hours ago added another item. So the the items are the first one is I'm gonna just give you an update on the preparation for the hacktoberfest?

A

um The second item is suggested by jonas about the update on where he stands, with his work on the templates for the nuts for nuts and uh the uh his work on the binding definition for nuts, and the last will be uh what paul wrote. I'm asking about the progress and um how the discussion, how we will continue with the discussion about the view property. It's a for those who don't know what it's about.

A

It's a issue created by michael from solas a few months ago about adding av property to the to the spec, to address the confusion, around publish and subscribe operation.

A

So the the first topic oktoberfest, so the octoberfest, starts third, on thursday, first of october, um at the moment, I'm ready with a blog post um that introduces our how we participate.

A

In the event, um it's not yet reviewed and approved, but it's um I I have it I'm just waiting for for a much better graphic that fran will work on for me because I I suck at graphics. um um So that's part that part is done um so the like written communication, the um I also finished working on the set of issues for oktoberfest participants, we're going to have around 75 issues with different difficulty.

A

Most of them, of course, are trivial and easy, but there are some some moderate as well or more demanding so, and it's not only coding it's stocks and ci, so I think we can attract many different people. um Let's take part in this event, what left is a set of videos that I will start recording after this meeting, uh because I want to make like onboarding to the uh to the community and the the event as easy as possible. So I'm not.

A

I'm gonna have a short video about asking api short video about um oktoberfest and why we are joining. There will be also a video about all the issues that we have I'll. Just explain how to understand like how easy are the easy ones.

A

How difficult are the difficult ones we already agreed also with run on the schedule for the office hours, because we want to make it as easy as possible for people to like interact with us in case they need some support with the issues or they want to do maybe even per programming remote programming, um so we're gonna host a office hours twice a week uh through in throughout the entire uh october.

A

So that's also done um and yeah and then also there's gonna, be a video about how to quickly um fork a repo and like set up a um set up a work with the with the repo and that's it so tuesday, wednesday left for the videos and and finishing out the stuff. So I think we're gonna make it to kick off on on the 1st of october.

A

B

No question question: I'm glad to hear there's some tasks I can help with because I'm not a coder right. So that's good. There will be.

C

I'm actually like this is this is great, actually uh and and thanks for bringing this to to the agenda of facing kpi right like to work or to participate on octoberfest. I think this is great.

A

Thank you um so jonas, would you like.

D

To continue uh yeah sure.

D

I don't know what you.

E

Want to share screen or.

D

Yes, it's a good question, I'm not sure if we get anything out of just showing the screen, though, but I can do it. Okay,.

A

Then I'll give you a co-host.

A

Can you do you have access to screen sharing or nets disabled okay, the front? Because I don't when I click on jonas, I don't have this option like.

C

A

Have the option now yeah, okay,.

C

D

Go away: okay, nice, all right! So for the typescript uh template, it's basically done as a stand right now I did find a few more box after I requested a review, um but.

D

But that should have been fixed now um and that's basically, the same with the.

D

With the net version.

D

I still have to lay out some more in the readme file and what the pr should solve and what's missing.

D

But other than that, it's more or less ready. uh I do have some questions in regards to um in the net version on the net template. It needs some binaries to operate fully and I'm unsure, if it's possible to if these binary files of the net client should be included when generated or if it's even possible, to do something like um like dependencies and note.

D

um So I'm a bit unsure how to do that um and I'm yeah. That requires a bit of research before I I can conclude what's best um and can.

C

I ask you what this binaries are for.

D

It's for the like. When um for the typescript version you just declare which dependencies you want and then you just npm install and then it takes care of everything, but in it it's a little bit different because, as I can see it it all, it needs to have the binaries in place. It can't like do the same process.

F

D

There are packages.

F

There are package managers for this well microsoft. Microsoft have kind of inevitably been promoting this ecosystem for a while. um I can send you around some links because.

G

Yes, because, like there.

F

Are there are places now you can publish decent.net ones so, as per always, nothing is as easy as it is of js, but I have to be good to get with the standard practice on dot net.

D

Perfect yeah yeah! Please do because that's one of the issues. I have at the moment um figuring out how to do these dependencies yeah.

F

It's a bit more like maven or gradle, with java land more than npm, so you have to actually run a process yourself, but you can just pull them. Okay, perfect,.

D

So I can just declare them and then, when, um when the when the other users uh open these generated clients, they install it basically right.

F

D

F

Yes, there are, we have, we can basically commit to using certain dependency managers and or, and then just like submodulate, if necessary, okay, perfect but like it does depend which ones we're actually willing to use. Also hi, uh I'm ben.

F

Are you meaning to sign up for these? But these this is a clash with one of my team meetings, but I'm now gonna try and turn up more.

C

Often I was, I was wondering, jonas, maybe you want to share this. This link to the pull request on the chat, so yeah, so bank can have a look at it as well. I mean ben and other people.

F

Yes, tom from my team should also be joining these uh more often, I basically said no. This is now time you have a side to go to all these meetings, because they're important.

D

So yeah just yet the link for the.net pull request. Thank you very much so yeah yeah perfect. So that's one of the issues out of the way.

D

What is left so for the bindings?

D

um I mean it's pretty temporary at the moment, because we're still waiting on um not stages updates but for the request, recruit, request reply um feature for the spec um to be solved somehow um so at the moment, it's temporary with this, this type of channel, whether it's a request for player pops up.

D

And then I use this request to apply channel binding to specify whether it should operate as a request or reply basically and but yeah. This is temporary until we get the the other issue solved.

D

C

By temporary you mean that this is already working, but this is what yeah will probably be removed in the future right.

D

Exactly yeah, okay, so temporary in the means of we're still waiting on some other issues to be solved before we can actually do this probably possible correctly.

C

But how how do you feel about um releasing this binding? As version 1.0? I guess yeah exactly I'm saying that uh yeah? Is it? Isn't it a little bit premature? Maybe, like probably, you want to have something like zero point, zero point, something and test it a little bit more. I mean let people test it a little bit more and then it's okay.

C

If we, if it's stable, we can just release, make a make another change to the version only without changing anything else, yeah all right and because, if if we couple these changes to the version as well, if something goes wrong, then we'll have to release version two right. So.

F

Yeah, that's why yeah brief.

C

F

I missed start of this. I do apologize for that uh nats. What is what what which particular acronym we're using for that is that's responding to in this case, uh because I'm I'm going back to traversal things right now in my mind, so so.

D

That's something: yeah nets is uh basically just an uh what's. It called it's an um um it's a protocol in the same base as uh kafka or mqt.

F

Okay, cool, so that's unfortunate, naming uh you know, nats traversal is when you go through someone's router and to get to a different subnet thing. So, every time I see it on a networking thing, I'm like wait. What so I was looking.

B

At it, I see no traversal.

F

Mechanics here.

D

D

Yeah there was some you had some comments uh friend about the typescript in regards to the wildcard use.

C

Yeah um the thing is that if I'm not, if I remember well, let me just open this in my computer, because I can't I can't really see.

C

C

C

Then this wasn't the typescript one.

C

C

C

Yeah, so what I mean here is that you, you were defining like um the asterisk.

C

um It's a actually means a wild card right, but um we should be careful with this, because um the the channel name is actually a url right and it's possible to include an asterisk in the url right, so my url can contain asterisk and- and I might not mean a wild card- I might mean um a a a an asterisk and uh why I'm saying that is. I know that in that specific case, might not be that meaning.

C

But the idea behind the async api is that we can use the same file over time right with different protocols, so we don't have to redefine the same channel for now. This is.

F

Actually, a serious one, because this came out very hard in my use case recently.

C

F

Wild cards is the one thing I really mean to bring up, because we need a uh kind of a an async specification of a wild card character which can then be translated per protocol because.

C

One, so we do have it, we we have. The the parameters is actually that right. uh So having like, I was, I was commenting on this uh issue. uh Maybe you want to join make. Maybe you want to do the fire, the font size a little bit bigger.

D

C

F

Better, oh that's much better!.

C

Yes yeah, so so the idea is that I was saying that maybe you can some people if, if people want to use people want to use a wildcard, you can use a parameter right which is called can be called rest or icon or whatever you want to call it right, like the rest of the channel right or whatever, and also how, though, how does a wild card apply to the public's operation, for instance right?

C

So if I'm publishing a message not to the public's operation, but you know what one publishing right when you want to send a message to a channel if the channel contains contains a wildcard. How does that work? Is it going to publish to all the channels that match this wild card, uh this yeah this wild card, or is it not gonna work because, like in in in some systems like, for instance, in gravity and q, with a nqb and all this stuff?

C

You also and also mqtt, you can have wildcards, but only for subscribing right uh when you want to publish. If you include a wildcard, it will not mean what you think it will do right like it will not be used as a welcome.

C

You know what I mean yeah, so so that's why I was saying that maybe that that's why we introduce parameters, so you can define, uh for instance, default values, um schema that it will be matching and uh and all these things right that maybe if you think it makes sense, something that we can add on spec in the future and it kind of it was it was there, but I removed it because it wasn't clear before I launched uh version two I have on the parameters you could specify what type of wildcard it was, but you could say if it's a wild card, if it's a wild card that spans among uh across different uh levels of the of the channel name like levels understood, as um you know like.

C

um I can explain this like levels. Are. You know that the channel names are urls and they are divided by slash right. So let's say that each slash is a new level of the the channel name. So how long like? How? How long does it spend the wild card? Is it only one level? Is it multi-level?

C

F

C

A separate wild card per level then exactly because and also because you may not know in advance like you want to have um you want to subscribe to all the um all the channel names. That's done like this, but I don't know what so.

F

We know so we don't actually know if their hierarchical is not is the problem, we don't know if they actually come from a tree or if they're, just flat.

C

Correct so in in in in protocols who are uh who who have uh hierarchical topics, you can apply it in protocols who don't uh that? Don't don't have that and you replace the slats with whatever else or you just lift this last, but it it will mean exactly that, that's less, it does not have any meaning right, so it was removed because of that it was, it wasn't clear.

C

It was, uh and and now that you're adding this to this specific uh use case, if you think of, if you think about it, you are in what you're doing here is that you're interpreting the spec in a specific in a special way? Just for this uh template right yeah.

C

If tomorrow I want to switch from typescript nats to nodejs mqtt, the the sync api files will remain the same, at least whatever is not in the bindings will remain the same like the bindings can change. Of course it's specific to the protocol, but the rest will remain the same, and this will be and so mean the same across any language or or protocol right.

C

You know what I mean.

D

Yeah yeah, that makes sense.

C

um Because it's it's me, it's mainly that and and the other. The other use case is that we want to have a single async api file for two different channels: uh two different communication channels, not channels as indicating kpi. Two different protocols like I want to publish an http api public api and people can subscribe via servers and events, and I also offer a kafka channel right.

C

So people can subscribe to the same to the same api via different protocols and or one can be nats and the other one can be uh servers and events and um the core of the specs will be the same for both right, otherwise we're we are starting to tweak the spec on for specific use cases and can become a mess at the time. Yeah.

D

Okay, yeah, I see what you mean so yeah, I'm just gonna keep this um uh with the parameters basically, and then I'm gonna remove this, maybe add another section about why we do it with parameters but.

C

Yeah yeah, if, if you really need that you're missing something like you need to do something that is not possible, yet, let's, let's create a a miss on the spec, so we can improve the spec over time. Yeah.

D

Okay, I don't know if there was anything else.

D

Hey lucas did you, I can't remember if you implemented anything in regards to boolean.

D

Template parameters, so it's possible to just do.

D

This, oh, if you can see.

A

You mean in generator or.

D

Yeah exactly right now instead, but you have to like do this now, you still have to do.

A

This like explicitly right through yeah,.

C

Okay, oh this is the boolean parameters.

D

um This is something we want in long term, I'm not even sure it's possible with the firm.

C

We we had the conversation for that right like um that. Maybe this can be specified on the packaging of the template. You have a list of parameters right and um would be great to add the the type of the parameter there right. You know what I mean.

F

Just a quick question because it totally just came to mind something. I was testing something earlier this week last week. In fact, so do we have a way to specify the uh size of said, pool or sediment what sir the size of 64-bit versus 32-bit uh it mattered, because I was trying to read on some microcontroller and I started having. I was not sure what size of ins or bulls I was getting out. The other side.

F

I'm not sure I understand.

C

What you made so so.

F

So if I so, the spec specifies a certain parameter as an integer right and with mpgtt is just a binary blob, I'm getting back so I need to so. By my end, how do I know how if it's a two-bit, integer or 64-bit literature, yeah.

C

Just just one sec then, because I think uh jennifer isn't yeah. He didn't finish with uh sorry.

F

It's an unrelated thing.

F

No, it's not really it's just. I saw boolean and I was like wait. What size of boolean.

A

I know so it's uh the topic we're now discussing it's a like the generator that we use for generating docs or code.

F

Oh no, where is that noise.

A

Cli allows you to to pass parameters and you can specify different values for those parameters.

F

Okay, sorry, I must have missed a little bit. I do apologize.

A

um So yeah there's an open issue on generator, so we had a discussion on on slack and it's a there's, an open issue on on generator that yeah. We should definitely improve the things about how we define parameters in the in the configuration, so we can even specify a type or even we mentioned that it would be even nice that you can specify schema but yeah. That's not yet uh doable with generator right so yeah. So that's the thing you in cli. You have only string right.

A

So then, then, um but then the problem is that you can use the generator also as a library. So you don't pass strings. So that's a that's a tricky tricky thing, so we first need to somehow introduce those types that you specify explicitly what type of uh of data you expect and then we can do some magic and now, unfortunately, you have to do the equals. True.

H

C

All right, perfect, um yeah, actually, actually the cli will be doing this magic right right, yeah.

A

But it has to have this information right.

C

Exactly to know if it will be casting to to bowl the end or not,.

D

All right, I don't think, there's any other questions in regards to this. Unless you have anything any of you, so this should be changed to 0.1 in the bindings and then I remove this parameter uh asterisk.

C

0.2 right because 0.1 was already there right.

D

That hasn't been released yet, though,.

C

Well, I mean we don't have releases and binding so so yeah my release, I mean you know tagging. This is zero. Point tagging, not tagging like uh adding the version number there right. We don't have released some bindings yet so.

H

So yeah shouldn't this just be 0.1 or.

C

It was 0.1 right.

D

Oh yeah, true yeah. Okay, even though there's.

H

C

Yeah exactly so 0.1 was there's nothing 0.2, there's something right.

I

C

That's right, that's what I was saying.

G

Yeah, that's all I got. I think. Okay thanks man.

A

Yeah so we're merging those right like we're, merging the typescript and if it's ready, yeah.

C

D

Yeah, it all depends on how, um if you want testing for it first or if you want like how sure do you want to be that.

H

It actually works in every case. Basically, so honestly,.

C

If you think this is good enough, I will I mean the templates. um I will go for it. I will merge them because the bigger this pull request gets the the less people will will uh have the mood to to look at it right yeah, so I will just merge it and if something fails, we'll start the use of creating issues and fix them right. Otherwise, this is a big ball of math yeah that nobody wants to look at right.

H

Yeah sure so yeah and we will anyway release.

A

It as zero one right.

H

A

H

So don't worry, it's just.

A

Experimental, did you add the release pipelines the github.

D

Actions too, oh.

A

Yeah, I actually did.

D

I actually did that yesterday for both templates. um So all it's missing is basically the uh security keys. Okay um for the release.

C

Will then will then the feet with exclamation mark will create a 1.0.

A

Yes, it will yeah. Oh exclamation mark means it's a major release yeah, but it's a first release anyway, and I'm not sure if by default it's not going to do one zero, so the safest would be. If you mark those two initial requests as hor c-h-o-r.

A

E

I didn't pronounce it in the wrong way. I was actually thinking.

G

E

G

Gonna, do it when you gotta, do the emerge yeah, but.

A

You can do.

G

A

Okay, so we're just um the initial release, I think we're just gonna do manually, it's me or from probably me yeah.

D

I mean yeah, let me just I'm gonna, take you as soon as I'm. I'm just gonna. Look through it real quick before the final final release and then I'll.

A

Just pick me when you're ready yeah, because we also need to add those um topics to the repo. So it's visible on the list of the templates also update the the generator with me with the list of the official templates yep, but yeah.

D

So how can I get back the.

A

A

What do you want to do.

D

I want to get back.

H

Oh I'm just co-hosting. Where is that.

C

H

I already did it.

C

A

um So defensive, are we going now with something from paul or ben? Do you want to like finish your question that you had about those the integer size.

F

Oh, um that's kind of like a brief question. I suppose sorry it was. It was a bit every now and again. You know your brain latches on to something you go. Must ask that question now, because it's been burning in my brain for a few weeks. um It's supposed to go for a brief one. It's like particularly with the micro kind of the microcontroller use case.

F

It becomes quite imperative to you, know, know what types we're sending around, because implicit implicitly like people tend to use, particularly with these things, as you get 64-bit, integers 64-bit floats out of like those libraries on pcs these days, just because of native types, uh particularly js, because everything's a 64-bit double. It's all double behind the scenes. So when you start sending them over the wire to a microcontroller, particularly over amputee, because it's just a binary blob, uh it becomes imperative to know the actual bit width of what you're sending so like do.

F

We have a decent way to speak because we have types already and they're really useful, but it's it's a bit hard when you don't when you're, you know in a c and you don't want to have bits over running so so you mean on the other message right on the message: definition yeah. So on the actual spec definition. Like so say, I'm.

C

F

C

So just re real quick because that's because that's unrelated to to very.

F

Quickly, yeah sorry yeah.

C

um So what we're doing on the schemas and on the async api is we're using json schema. So, whatever is possible in this schema, it's uh it's possible on this api.

F

Okay, so it's just it's just.

C

um So if you look at the at the json schema specification, you will see there are ways to specify this. There are ways to specify numeric types, let's say and we're using version. 7.0 of this schema and uh there's ways.

C

Sorry, it's not here is actually on the format. I think suggestion schema format, um uh numbers.

C

Yeah, let me just uh briefly check for it and.

C

Can't find them now, but if I, if I find it I'll I'll, let you know, but um there are ways to I think, or I remember there are ways in this scheme to specify um the type of number the type of number it is right.

C

So you want to you want to know exactly what type of integer it is right.

F

That's pretty much it I wanted to say assigned unsigned or something else. So there's there's this.

C

I'll have to check, because I think distance schema eight, which is not. We call it eight, but it's not. Okay,.

F

C

It uh well I'm distracted.

F

Enough so I'll take this offline.

C

Yeah, let's, let's just chat about it offline, maybe in the slack channel and uh I'm saying because yeah uh you know paul- is waiting for uh the topic he he brought to you.

F

C

This is a distraction. I'll tell you um yes, maybe paul. You said that.

C

C

Instead of cost.

C

This is more power, that's.

H

Cool yeah- that's not always cool.

E

With great power comes great responsibility, exactly.

C

Yeah so comment from paul is that it's about the view property right so.

C

Let me just go to the issue love to make just a point here that, just as a reminder that discussion is still open, but I guess.

C

I guess the discussion is not so much about adding or not adding the view property, but how do we solve this public subscribe um confusion right? That's confusing many people right.

B

So and that's what prompted my question, I was chatting to a customer and they had it completely the wrong way around and I had to go no.

I

That's not what.

C

I

C

This this is confusing many people say even myself. Sometimes when I'm writing the specs like, oh, no, it's the other way around so depending on what I'm doing. If I'm writing the spec uh from the consumer, let's say consumer point of view like it's not my api, I didn't create that. uh Then it's easy for me to think about it, like what can I do with this right?

C

uh But if, if I'm the author of and and I haven't, created the the code yet and um and I want to write the spec to generate the code, for instance, then then it's like you know uh your your brain is completely flips and it's like uh what doesn't make sense at all right. So yeah, it's it's! It's tricky, it's something that we have to solve, but um I added here like uh explanation why? I think this is um I'm gonna. This is not a a good idea.

C

I mean it's a good idea as a starting point to discussion, but why I think this is gonna, introduce more um confusion right, like this view, provider or view client, uh and it's because of the side effects right. The side effects that this is gonna create on the rest of the spec right, because we are adding a flag here that is not constrained to the info object. It's a flag that we add here. That is going to condition how we see the rest of the same spec and especially the bindings.

C

So one of the examples that I find pretty often is, for instance, with the http binding, depending on this value.

C

When we have the http binding on a channel or on a publish operation, for instance, sometimes it will mean a request. Sometimes it will mean a response, depending on the on the value of the view flag right so yeah, and uh we that's that will be coupling the definition of the http binding to the value of this of this flag right, which is completely and related to the http binding right or any other binding okay, so so yeah I was.

C

I was elaborating here a little bit more like what's the what's the purpose here in the end, like the the the I know that the the issue is called add view property to the info section to change the perspective of subscribing publish operation, but I I'm like I'm thankful for for michael that he created this this issue.

C

That's uh that's creating the a great discussion, but this issue title itself: it's implying a problem and a solution right together, like we have this confusion with publish and subscribe point of view and here's the solution, which is great because he's proposing a solution right. um The thing is that I think now the real purpose of this issue is more on. How do we solve this?

C

How do we solve this because view is not really an option, because it's gonna add more uh confusion to the rest of to the rest of the spec to the other parts, so so how it is now.

C

I think the current state of this is: we need to come up with another idea on how to solve this right. um I've been playing a little bit playing, it's not the word, but yeah. I've been like trying to come up with a different solution, but that's gonna involve more than just adding a a flag, because I think- and I think I said it here- I think this problem is not really about publishing subscribe.

C

I think this problem is actually um reflecting a more intrinsic and deeper and intrinsic problem of the spec, like of the design of the spec, as it is right now and if I someday find time to to redo something that I did in the past and I completely lost. I don't know.

C

Why is that once I once dedicated some time to um to highlight the lines of the spec in different colors, depending on where they belong, like I'll, explain that so, for instance, real quick, for instance, this one this first line is: is this for the this is for the spec to know to know which version it is, but, for instance, this is information about the application right. The application not not meaning the api, but the application, the code, the program right.

C

This is information about uh the api. It's not information about the application.

C

This is information. The verve publish or subscribe is information about the application. What the application is doing to this api to this interface? Let's say I'm subscribing to this or I'm publishing to this, but it's specific to this to this um application right to the user management service application.

C

If we go to another place to another application, another async api file, everything I mean, like everything, that's related to the api, like channels use the user register, even the description of the of the channel will remain the same, but the version but the sorry, the the birth will be, the opposite will be published or or maybe publish right.

C

So what I'm saying is that we are mixing we're mixing um and also if we, if we get deeper like message, is actually part of the of the api, not the application. So if you see this is application, sorry api application api. So it's all mixed together. So the the concepts of interfaces and behaviors of this of the application, what the application is doing to this interface, it's completely mixed and that's why it's impossible to break right. That's why you you have to go to another.

C

Whenever you have two apis, one is publishing and the other one is subscribing. That's why you can't reuse the whole the whole channel right. You can't, because the whole channel is containing the intent of this specific application in regards to this channel.

C

So so in my in my opinion, I think this requires to get uh to struct this intent if it's published or if it's subscribe, to extract it from the channel definition and put it somewhere else like somewhere else in the spec, probably on the root level or in a operations object or whatever.

C

But we need to first decouple these things right to be to be able to um to um to remove the coupling, let's say, and once we have the coupled, um we can then create different verbs. Like I want you to see, uh I want you to understand that I'm subscribing to this as a client or I'm publishing to this broker.

C

As a client- or I am the, uh I am the provider, if you want and I'm publishing to the broker or I'm subscribing to the broker right, um we can then move these definitions, this different verbs to another object, but that will require a new major version of the spec right, because it's breaking change so but again like, uh like, I said here, um I think uh somewhere somewhere in this thread, I may find it. It's like, like I said here, is that we shouldn't be afraid of releasing version three um of the spec and.

C

Yeah it's somewhere, I don't know where it is, but yeah it's here. It's actually yeah. So I think this yeah.

C

This problem is important enough um to create more than a band aid and uh it deserves an elegant solution that will fix it once and for all and um and it doesn't create more confusion and if we have to rinse version three we'll release version three right and we'll release all the tooling we need anyone will need to migrate from version 2 to version 3 right, we'll make it easy to to move to the new version.

C

Also we'll keep supporting version 2.. So there's no need to be afraid about that, but whoever whoever wants to start using the new version, a new way of defining these things, can be moving to version 3 right. So so that's I mean that's my opinion.

C

um I'm happy to I'm happy to reconsider this. If there's another solution, I couldn't think of another solution that will not create side effects like like the the view property right.

C

uh It's again, it's a complicated. um It's it's! It's a complex.

C

It's a complex problem to solve without really addressing the the intrinsic problem of the spec right now, which is like everything is super mixed right, um application or behavior and interface are completely mixed, and that's why it's it's super difficult to whatever you do, even if we change the meaning of publish and subscribe, there will be other people who will be confused right, so there will always be someone confused.

C

B

Yeah, no, no, I agree and it it's complex because it's simple right, it really should be simple, but it's not, which I think is the issue, um and I see what you're saying I I uh I think if they were decoupled. Obviously, then you could get some reuse from them. You could do a reference between them. I assume so that um that wouldn't be such an overhead.

B

uh It's just one of those things that when I speak to people the longer it goes on the more difficult it's going to be to get them to change their mentality and then unwind yep.

C

I agree so one kind of related to that um related and not related is we were thinking about uh establishing a um like um release cabins for the for the spec like, uh for instance, um making it mandatory for us to release a new version every six months independently, if it's major version, minor version or a patch or whatever like like every six months, people will have a new version of the spec and of course, we'll have I mean six months or whatever other amount of time we'll have to check.

C

But I don't want to commit to six months now, but um but like every every every x month, we release a new version and we adapt the tooling uh in advance and we make changes in the spec first and then we communicate what's what's going to be, the new version so tooling can also adapt other other other people's tooling can adapt and um and yeah.

C

So you know to keep moving the spec and fixing it because, um like exactly for what you said like the the longer it stays here, the the bigger the problem is right because it's confusing more people so we'll love to get this in the future. We'll have to get this type of problem solved in less than these x months right, um even if it means releasing a new major version like with json schema.

C

um You have decent schema draft four five, six, seven actually five now, but you have six seven eight uh as a last one, um and you still have people on on draft four, of course, but yeah for for those who want to do the the new cool stuff that you couldn't do with version four uh they can upgrade so so again it doesn't mean that we will not be maintaining this uh version two of the spec but yeah. We can keep moving forward right.

C

B

Okay thanks no, I I got. I spoke to a customer who was confused and also got an email sent. The issue had gone style, I saw it was reopened, but it's like ah okay. I wonder. What's going on with that issue, you know.

C

Yeah yeah, no, it's I can. I can tell you it's not stale we're just leaving like um like some decent amount of time for people to uh for people to jump in especially the summer summer, was in the middle uh to jump in and propose new solutions or propose other ways to fix it, and also because we are like full of work like we cannot handle everything right and um and also yeah.

C

So just to give some time, and once we have some once we left some time and once we have time ourselves to to work on it or someone wants to jump in and work on it and a different uh solution, um we're happy to do it and have to address it but yeah we we will be. We should be close to uh we're actually close to be working on this and this and many other many other stuff.

C

And when I say.

B

C

Like before the end of the year, so.

B

Hey you don't have to explain to me. I work for companies where priorities change all the time. Yeah it's like. No, I mean when is.

C

It being released.

B

It's like oh yeah,.

C

Yeah, just just just to make it clear for those who are watching the recording right. It's just that we will be addressing this problem, but yeah. This is a spec. um We need to be careful with whatever changes. We do right. You know sure, because it will.

A

Impact many people and anyway, this issue should be protected from the state board. So I guess you cannot, just the label keep open, and so this.

E

Input doesn't touch it because it's intentional that it's.

A

E

Take some time.

E

Oh, you don't yeah have.

C

Yeah, thanks for the watercolor.

C

So yeah, so I hope this provides some light uh on the status of this issue. um We're gonna solve it one day, we're gonna, think back and say, like hey, remember those times, but in the mid in the meantime, we'll have to gather opinions from different people right and.

C

Which? Actually? No that's? That's the pro. That's the problem. That's that's the same, meaning as in as in one zero. We didn't change it. I wanted to change it and then because we were, it was creating a lot of problem and I said like okay: let's be conservative, let's keep the same, meaning as one zero and yeah, but one I mean almost nobody was using one zero. So that's why we didn't have so many guys.

B

All right, well, thanks, I understand and I'll continue to educate the customers into the right way of using it. In the meantime, right.

C

In the meantime, yeah thanks for a little bit, uh but in the meantime the only the only way we have for that is just uh to tell people that uh look. This is the way it is now. We are all aware that this is a mess like this is super confusing and we're going to address it so yeah sooner than later, this is going to be fixed.

C

Thank you. There will be new problems.

C

Optimistic it's reality, I'm happy to have new problems every time it means it's moving.

A

Forward, okay, then one hour passed um all topics covered anything else in last, one minute.

A

So thanks a lot for joining and I'll, see you in two weeks or on our oktoberfest office hours in one week, yeah yeah enjoyed it.

B

H