 We're like the three musketeers Yeah Guten tag Thanks So has everybody recovered from coupon, I guess I still sleepy Me and Ricardo had to wake up like five in the morning those three days Two days. Oh, yeah, I had to work late. Oh, yeah starting early anyways because the company so Very long days for the whole week But your talk was very nice. I enjoy Thanks. Yeah Took a lot of preparation actually and if you I don't know if you've noticed I haven't said really much about facts in it. So it's really just I General dribble wish but yeah other the YouTube links out. I still haven't checked. I'm still on vacation technically. I Don't know. I just the links that we put up there on their website still work. So I just kept them That was interesting the access to the Google Drive was blocked. I think a week before cube con I wanted to still have a look at the other pre-recorded videos, but I couldn't I don't know the I think CNCF like changed their website completely. I think today went out and and I noticed the link to the Public events calendar was broken. So I had to ask for a new one but You know, I don't know what else changed or where anything is but the nice thing is like under projects you can go to all of the Sandbox projects and you'll see us there. Just we don't have a logo yet. Yes Yeah, just check the Playlists is not yet only the sponsor Videos, I guess we can wait for a couple more minutes. If anybody else is gonna be here If not, we'll just go through the agenda make it a short one. Oh, thanks for editing No problem, there's also a start did send out Kind of like to follow the governance thing that I pushed kind of an Email to the maintainers list manual for adding Ricardo is a maintainer and I think Just to kind of go by it Yeah, so that's all for you to kind of look at I am Wait a couple more minutes if that's okay Okay How are you? So Manuel should I share the screen and we can go from there or how you want to do this? Yep I'll see if I can press the share screen button here Go You guys see the screen Yeah, I can see it. All right, great. I just wanted before we get started. I just to show Doug was a little bit. He was like, why am I the default to signing for all the PRs? So hopefully there's a PR here that he just merged it did appear he merged it regarding changing the code owners file. So It's still could just code owners. I think when a PR just puts the default get help puts to default to signies So just to show we're not removing Doug from anything. It's just he's kind of like hey Why am I on all your PRs, which is fine? But anyways, welcome guys. Thank you. I hope you enjoyed KubeCon if you were attending Wait, oh Manu already updated everybody's great So we lost Ian but Okay, since it's your first time. Is it your first time? Do you want to be associated with any company? Oh Yeah, so I've been on the contributor this time with the Oracle Cloud infrastructure So he didn't catch it as Oracle Cloud infrastructure Oracle Oracle Small company, it's not All right, so I guess Community question time. Do you guys have any questions? Anything is cool So as far as the logo goes we're still waiting on CNCF recommendations. So Okay, I think you've commented on the great logos we got before I Put like an online logo generator last week and I I saw some recommendations, right? Some designs that look good I can suggest but I imagine those are probably copyrighted and whatnot and We'll probably need to have like somebody from CNCF Take a look at some like designs. Yeah, if you want I can send those over some that looked really neat to me awesome, yeah, that would be great because CNCF does have four projects and now that we're saying books projects There is an issue tracker and I open up an issue and it's like waiting for the design team So it's probably they're all busy because of cubecon last week. So or whenever but If it comes back, I will just link it to us on our slack channel to look at and then we can talk about But apparently this all came about not for me just letting you guys know this came out once we became sandbox Actually Chris Asian Chuck I came pronounce his last name, but he's a Lead of the Linux Foundation. He said that we needed a new He's the guy who created a repo for us and everything Helpful is and he said yeah, they came from there and then we went from there but anyways, so Okay, so let's see working towards Next release. I think we should start thinking about a 1-0 something. I think they will help with our People looking at us in a different light being to 0.1 see currently what we have on the old repository even It's okay, but we've done so many changes and you can look at the The roadmap file and see that we've worked a lot over the last since that And I think we're kind of ready. Yes, it is a more or less a marketing move, but we need one so So yeah, there is two things using and underneath. I have also released planning. So we'll get back to that But actually, let's let's let's go there since I started it So the one question is whether we want to use this version I propose some sort of 1-0-0 RC1 Manu also said hey, if we need a release, let's just do a 0-2 right manual and Then do a big bang on for cube con and a and I really like that idea I like that this cube con and a November 17th deadline But then I think that we don't really need a 0.2 until then it's just more work What do you guys? Then I haven't done much software versioning Usually worked with the small the minor bumps, but given that the major release also marks Backwards compatibility issues and I don't know if we really are backwards compatible Well with the specification it's different anyhow, right? We're not releasing a software product, but In a way something written in the current specification and the previous one they are there was some mismatch Or we dropped a few things. Yeah, there is some changes Definitely. So yeah, that's something to think about. I think what we have now is a lot closer to To something that I feel, you know, we should feel comfortable about having community actually use The only big thing that that I need to work on and I've been working on is the retry and error definitions. I think currently They're they're kind of like the weak point of what we have in many ways So I've been I've been updating that a lot but with that in I feel fairly comfortable of Saying hey and then maintaining what we have for the future as far as backwards compatibility goes I think we made a lot of good changes in the last month But again, I think I really like the manual your idea for Q code and a dead date Another question that we come is right now. We have to SD case as well, which we didn't have for 0.1 the Java and the go one and there is also an issue for open for What is it? the Microsoft guys asked for the SDK, but I don't know about that. So what do we do? I think cloud events releases them separately. So the SDKs have their own releases That may not really follow the specification releases So and even if the specification is on 1.0, let's say the SDKs have the right to have their own Independent versioning as far as what they want to do so I maybe we'll do it that way or How you guys feel about that Independent release makes sense to me Yeah, there's always gonna be some Kind of change or some like specific thing by tying it to the Specification it's gonna end up being a challenge Yeah, so it's good So it's we solve that by minor release versions Yeah, I think you do or you could you give an example of the cloud events SDK not following suit But they do denote their compatibility on the with me page They have like a table, you know a matrix of you know, like this SDK versions compatible with They have like the specification house, I don't know five versions and yeah, there's something Okay Pretty similar to this This is only the the cloud events SDK version the cloud events version So compatibility with the specification version 0.3 and See the go one because that's supposed to be the biggest one they have They don't have cloud event specification zero three one point. Okay So so their versioning seems independent and their release, but they just say, okay, which ones do you support? Yeah, so it's a major. Okay, we can do that. That makes sense That's good Okay, so we decided something for KubeCon and a and as the case are independent Good, I like that a quick question for KubeCon are we are we saying one or release or or is that still a little question? Like is it zero point two or one point? Oh, I Will prefer something with one point. Oh, maybe milestone a release candidate. So we'll have to talk about that But let's discuss keep discussed that is is that okay manual to keep this in Discussing this in the meetings and see, you know, if people have ideas for the next one and going forward Version, I mean what the version is it's not that bad is just we need something maybe a month before to come to an agreement So we can because there's gonna be some work on everybody to get it out Yeah, I wanted to mention that it might be a good idea to have sort of a Release time schedule. So if you want to go pump it to one dot zero that we leave maybe One community call that's two weeks or maybe two that makes it four weeks for Buck fixes like a feature freeze, you know, and Release candidate is a good idea. I think that release candidate one typically also includes the feature freeze But again, we're not doing software product. So I'm I'm good with anything And then the idea was having it closer to Kube Connors America Would make it a nice marketing Gag maybe Maybe we could even have a blog post CNCF, I don't know Cloud events had that I'll be nice. Definitely. Yes. Yeah We should probably engage like if there are any marketing contacts that are provided by CNCF just to make sure we talk to the GTM plan and talk through all these like channels that we can get Get this announced through Yeah, we really need yeah that we need more of like community engagement and stuff like that. So any help is much appreciated So it could also be in October it doesn't need to be that close to Kube Connors America I was just wondering if we should Really do it now if now is the best time or if we It's probably not worth making another version in between so Today to zero point two and then are going to one point zero later So maybe something in late September beginning of October would also be fitting, right? So if we have the release candidate now and then Work with bug fixing towards the one point zero big release including all the marketing Okay, yeah sounds good. Yeah We want to now go Some of the things that changed since our last meeting it wasn't much it was a week of waking up super early in between for this conference and stuff but we did update the for each and parallel states And What how can I just this is a big PR, but I'll try to explain And I think we did we touched on this last meeting as well but basically for each imperial states had used to have a set of nested states that defined the control for logic for The parallel iterations remember both parallel and and for each states Execute in parallel, but the for each one each of the quote unquote branches gets its own Element of the iterating array where parallel states each of the branches get the whole States data the same copy of it, but anyways Long story short that that was so ugly I think that having nested in states and sites branches of parallel state or for each state Was too much especially when we have the sub flow state So what we changed is that instead of defining and then you had also sorry rules like States inside of the branch cannot transition to states outside the The branch and blah blah blah So we simplified that where we just said okay look free if you want to execute control for logic Inside of each branch you specify a workflow ID The maps to a different workflow where you can do that. So for this case you need multiple J jasonry ammo files, but for most Probably usages for for each imperial states. We also added the ability to do actions instead of states so so that I think is a good change because we would really it drives kind of simplicity of these states a lot and Does not really Enforce some rules. They're kind of silly such as the transitioning and stuff like that Anyways for the event states I did this and I hope you know it would change the on events action at the events actions Which really never made any sense to me to just saying on events so basically we're saying is The event states the on events defines What events are you looking for and then what actions you're going to be executed? Depending on the events that are being consumed. So just a small change there Okay for switch state and we can allow transition or end the definition. So this was a big one that actually I'm glad we did is Switch states defines a one or more conditions and each condition if it evaluates to true can have a transition to a different Part of your workflow or the or different state if none of the conditions match you have the default property Which will then be the default transition kind of like similar to X or gateways. They work in BPM and two for example One thing that we were missing is let's say if a condition matches you want to stop workflow execution in this state in that case Since we don't have an explicit end state. We will have to kind of users would have to add a state There's just an end state So we change that where both in the condition Expression and in the default you can either define a transition to a different state or you can define straight in there An end definition saying okay if this we want actually to end work on execution So that was the the motivation behind that And the last change that we did was for transition and end definitions and in our transitions You can actually produce an event the same thing with end definitions. You can have a End of end definition with a kind Produce event, but we only allow the single reference. So there were cases that Actually Ricardo was working with one where he's saying hey, dude. I want to produce Multiple events here not just one and we looked at the specification. I said we can't do that right now so We added that where it's now in an array of event references So you can define on multiple Events to be produced during those two constructs, so that's kind of like What we have done and just for my if you guys know like for our issues here the one that I'm currently Working on is this guy basically just updates the retries and the on-air handling for specifically looking at timeouts right now we Don't really have a clear understanding from what I've seen how how that timeouts are related to errors or retries We kind of assume that that just happens, but So updating that and I think Ricardo is also looking at this guy, which is he proposed You know additional parameters and how they're asked and there was the Vang I think he's from Ali Baba and He pointed us to to some specification docs to see how we can they can help us out. I'll look through that I wasn't really He's asking for an awesome kind of a specification or RFC to find how we can classify the Parameters, okay, I Sumer he Has the same problem on any ballah. So, yeah, the problem here is this we actually we discussed that on the last meeting if I'm not wrong and Be sorry. Yeah, no, no No, that's the Actually to be honest I I need to take a look on this again and Maybe formulate it a little bit more You know for others to understand what we are trying to achieve here That is having a classifier for the functions in the parameters So users would like oh This in this function. I'd like to send Parameters this way, you know because If you if you think on functions, maybe you can have a mean parameter on now parameter in and out parameter, you know, I mean When I used to program that Cpo procedures that you have like Classifiers for parameters actually on it. It is not just input parameters, but maybe you need to classify the return, you know So that I should be honest, I don't know how to tackle this actually I maybe I need to spend a little more time thinking about it and maybe consider adding a I don't have a draft or design doc of something that we'd like to implement with this pack and then It would be nice to understand that if that would make sense or not for this case particular we are trying to achieve like a function that make make mix press calls so For instance, I have Parameters as input, but I'm sorry Path parameters. I can have query string parameters. I can have body parameters Or I can send parameters as headers and then how can I express that in 10? with this specification And that's important. Yeah, and then that's that that's the case to be honest I don't know how to express it on this back, but I know that I need that We'll figure it out Sorry Other than this Can we have the the extension issues those are a long term I haven't given a much thought about either if anybody would like to Tackle those Be happy to The net SDK that's something that we got asked for By I think an IBM guy, it's a good idea We just I don't think we have anybody currently that's proficient in dot net to tackle that but It would be nice to Actually, we don't have we don't have time right now. I at least I don't have I would like to create this But right now I need actually to update this they go SDK with the latest Spec and also add some features in there that I'm missing. I hope that I can get that So I really we realize that they want to zero that would be parking Yeah, the only other thing that I have that might we probably have looked at in the Java SDK I We also added Validation so not only can you build Your workflows programmatically or parsing from the Jason or YAML let me find the section Yes, you can also have workflow validation where it will give you schema related Errors, so if you load up your Jason or YAML, it will straight up tell you hey, this does not follow the schema Or if you create your workflow either or if everything part parses fine It will it will give you extra kind of business errors for example Hey, you don't have an end state and definition defined in a state you you have multiple start definitions defined and Stuff like that. So that that I think can we do this in go or I don't even know I don't know whatever, but this is just an idea of something to add it later. So yeah, we can do that Okay Um Come back so that's kind of all I have I mean manual or anybody you guys have any other things to bring up Don't have any additional issues, but I think maybe we should consider Selecting what we want to get into the release and what we want to work on So our handling. Yes How about those extensions that the tracing extension? That's our extensions always are independent from the main specification on day Yes, the way we did them is they're complete separate Jason files and we have this extensions page here, which currently we just have the the KPI extension And we should we can add them here independent and think of releases, but maybe here we should still put down What versions they're compatible with but the way we kind of did this is they all reference States by ID or name as long as we don't remove those two properties on states and work clothes We should be compatible to all, you know, many different versions in the future But yeah, that's a that's a thing We I would like to have hold on I go to issues for the tracing one and I think Ricardo linked the open telemetry spec It would be nice, you know, if we had or work with Some, you know, somebody who who is an expert in this area To make sure that what we're doing is right. So I'll try, you know, maybe somebody even a red hat We can pull in and ask them, you know, if they can help us here What do you think Ricardo? Maybe or if you guys know anybody, you know, your locations that Maybe guys from primitive steam they they have more, you know experience on this Or from the Yeager, maybe we have both people working on the on those projects. So yeah, I guess maybe Yeager. I know a guy named is Juden see something like that. I can recall exactly, but he is in Germany, but he's Brazilian And he works in the Yeager operator. I can you know, I can talk with him Yeah, we gotta help as much appreciate it or just to make sure Cool, thank you Anything else you guys want to talk about any questions anything is fine And for me Right in that case, I guess we're done and we'll have our other meeting in two weeks and let's think about, you know I guess the most important thing right now is the Release and what we want to call him and we'll discuss that also next week If you would like to I would like to you know get some more communication with the air handling and and the timeouts If you have interest in that, please let me know. I'd love to kind of do it like a community effort on that If possible, so we we all kind of discuss what you maybe take a look their Specification is an air handling section That that shows how to do on air and and retry definitions And and think about like how right now there is no correlation between a timeout, which is a date and time definition To the way AWS does it is similar to what we have However, they have a specific technical errors so on on air they define like if air Name is timeout air or something like that and we don't really have that. So that's the disconnect right now that A timeout definition should either itself define retries Or issue define air handling, you know Just be a timeout as far as value goes and then have the owner and timeout separately But I'll show examples and stuff as well when I Wanted when we tackle that So yeah, if anybody would like to to help out with that, that's a big thing big help And thank you guys for joining. I mean really appreciate it One quick question the release planning we've just follow up on the issue right make a chat on the issue Okay Or Should we discuss on slack how to move forward with the release candidate and did you have a blood plan in mind? Which milestones should lead us to 1.0? Well, I mean this is saying like I think I have created a 0.2 milestone Just because this was a long time ago Because that's what I thought we would do But we should probably update it here if you look at the read me now. I think I mean the roadmap So this was our view one release in the old repository and this is all the work we have done This is a lot of work. We have done since then and We specify here even 0.2 At that time when we didn't know because a lot of things changed like we got three We went through three different repositories since then And and with the project status being now sandbox. I mean some things have changed But it would be nice manual if you want to update, you know the versions once we decide And then we can go maybe through everything that we have done And and I think everything should go in the next release, but you know Not sure. Um, what is it should go in the next release what I meant was, um, the dates and Like release candidate locking feature the this sort of past that leads us to 1.2 We can discuss on the issue. Yeah, yeah, definitely, but but the big picture is is more or less like We need some sort of Not only to show progress which we are doing Anyways, we're always progressing, but it's to show the community progress and and then what's the best way to do that? I think it's as far as versioning goes to have some sort of major release Um, I'd really I wouldn't say this you know if I didn't think that it was feasible at this point Yeah, we're not perfect and yeah, we're always going to have issues and grow But finally after a long time working on this I really feel comfortable with small changes and updates where I can say hey, this is actually very usable Um for the community and it's just allowing them to to see the same You know, and I think a major version is is a big part of that. Maybe I'm wrong. You guys tell me. This is not a monologue um, but I've yeah anyways, I wouldn't want to download an SDK version that um goes on Specification that is still somewhat in the development branch and doesn't have a a number on it Well, yeah, specifically, you know go there with sdk's And I think with go is the same and we can you can speak for it, but right now since our version is uh snapshot We and I guess I'm putting on zero two already We only released to the snapshot may even a repository. So once we put an actual release of the SDK will be also on the main Yeah, maven central repose and then it's it people will feel comfortable like you said manual to use it. That's right Yeah for go it is like uh when we really do a tag release um, it is already on already for others to use because uh go used like this github Way of uh The paintings management. So you just say on your go project that you'd like to use these Projects from github and you specify a tag that you would like to use and that's a go wheel Download the stair the sources and then you can do just So we'll be ready for that Okay We don't need a third party repo like nexus or maven Yeah, okay regarding the anything else. I don't know if you've noticed but um denkorn may uh merged my Pull request on the landscape, uh, which changed the name of the project card in the landscape from serverless workflow specification to just serverless workflows Oh, nice. Nice. Um, yeah, I think that was also in the initial Proposal to the toc and somewhere along the way it became serverless workflow specification Because that was what we were doing that we didn't have sdk spec then right? so, yeah, and yeah, okay, I think we Michael you joined us Michael are you there? I am I am um, I only just recently become aware of some of the stuff you guys are doing So I just thought I would join to um, see some of the the meeting comments and I whatnot, but um, yeah, hi That's nice. Welcome Thank you For attendance. Do you want to be associated with any company? Um, I can't be oh, I'd send that on at this point. Um Uh, yeah, you came for Morgan Stanley. That's fine The partner didn't catch it. Uh, which company is it? Morgan Stanley. Oh right So do you have any concluding words for today's community call? Uh, no, I think I talked enough already But again, thank you guys for joining and I mean see you guys in two weeks and yeah, that's all. Thanks All right, thanks everybody Cheers