Layer5 Meshery Development, 17 Aug 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Development Meeting (August 17th, 2022)

Description

Meshery Development Meeting (August 17th, 2022)

Agenda:
[Pranav] Progress on Meshery Dashboard PR#6038
Bold lettering in column title? (Lee)
A bit too much whitespace between the "Workloads" heading and the table, perhaps? @Ashish mentioned a Namespace selector - that makes much sense to me. (Lee)
Left alignment? (Christopher)
Icons for resources? (Rohith)
+1 - Kubernetes icons and their colorized shapes have begun to emerge in the /ui folder. (Lee)
[Chris k ] Minimizing the number of files in the root directory of Meshery PR#5837
[Ashish] Adding user friendly API on capability registry PR#6050
For thought: Move /components under /model/components (Lee)
For thought: Confusion over version in endpoint. Clients are not seeking version first. Clients are seeking categories, types,... first. Consider version as a query modifier (a filter).
[Gaurav] Updates on the card view of Application and Designs #6046
Make grid view responsive by auto-hiding button labels. (Lee)

[Ruturaj/Aditya.c/Uzair] Updates on Meshery Catalog capability PR#6028 PR#6027
Let’s agree upon Clone, not Fork and use a Clone (duplicate / copy ) icon. (Lee)
Hide Save and Delete actions on Designs not owned by the user. (Lee)
Add authorization checks to deletes
[Yash] Meshery-ui-restructuring branch updates Open PRs Tracker
[Harkirat] Creation of reusable cards for ui-resturucturing #6051
[Harkirat] Docker Extension approach for #5782
[Nishant] Spec for `meshery app import` command
[Aaditya] Components on Meshery Doc
[Aaditya] Compatibility Matrix and its Upliftment

Join the community at https://layer5.io/community

Find Meshery on:
GitHub: https://github.com/meshery
Twitter: https://twitter.com/mesheryio
LinkedIn: https://www.linkedin.com/showcase/meshery/
Docker Hub: https://hub.docker.com/u/layer5/

A

Yeah so uh welcome once again uh in today's uh mystery development meeting and we are discussing about the machine development plan on wednesday. uh So with convention like we start with uh uh introducing getting introduced to some new commerce uh on the call getting to getting us, know better and uh letting them know us better. So uh I can see.

A

I can see gabriel, uh I think, he's uh someone who has joined the call for the first time right, gabriel,.

B

Yeah good evening, good day, yeah.

A

So if you can tell us a bit about yourself nick all right, so I.

B

Was at the last meeting I was last week, thursday. I think yeah.

C

B

C

Yeah gabriel thanks good of you to jump onto the measuring call as well. If you haven't introduced on this call now's your chance, um there are different people who show up to the four or five different calls that we have, and so it's always good to get acquainted.

C

um Would you like to say hi just.

B

Okay, hi everybody, I'm a full-stack developer yeah. My name is gabriel again yeah, so I joined mercedes through a friend. He introduced me and asked me if I wanted to participate in an open source contribution, so I said yes, then he sends me the link, I filled the form I've joined the slack channel. I tried. I introduced myself already on the slack channels, so this is the. I saw um the meeting on the calendar that it was for today. So I joined to understand better commercials about.

A

Great great great thanks for joining, uh and uh I think uh deep has also joined uh it for the first time right.

B

A

So yeah, if you can tell me, tell us a bit like who I use. Where do you come from and your college yeah, but.

B

I'm a third year student uh third year student of computer science engineering.

B

I am a bit into open source and web development stuff.

B

A

Okay, I I think there is some network issue uh at your side, but we got you. uh You are an enthusiastic open source, developer and uh you're in the third year of your school, so yeah.

A

uh I think we can now start with uh our agenda items. I I see some zoom users here so if they can update their names, so that will be like really brief, so yeah cool.

C

Yeah, let me interrupt real quick because it's kind of a standing rule that we have, um if for anyone, who's on the call, um it's super great to have everybody. Please do us a favor, the kindness, if you would, of adding your first and your last name into the zoom participants list. So if sean is with us- and that's great he's been here before but ishaan, if you'd asked add your last name, that would be nice.

A

Because of my exams, but I'm back.

C

Terrible nobody see the thing is is like nobody needs those exams anyway, so, like you, can just set them aside and just forget about them. Come over here participate. Now I totally kidding um good. We had a um someone on the call as zoom user, and so, if you didn't get a chance to introduce, thank you for updating your name, but anybody else that didn't get a chance to say. Hi.

C

All right, avi, thanks for letting me interrupt. I just wanted to make sure we got everybody.

B

Yeah thanks thanks.

A

Okay, so great great great thanks for your introduction, uh let's move on so uh the first item on the agenda list is uh progress on machine dashboard uh software now.

D

So the ask is that, like, whenever you log into mystery, um you see the dashboard, which does a pretty good job currently of telling us the connections which gives you a chance to get the connection status to your doctor to your connected community special.

D

If you're running any adapters, you can check whether that, whether you are successfully connected to them or you can also see whether your prometheus and grafana are connected properly and if, in case you are running service messages like me like right now, I'm running used to service mesh deployed by mentally, so you could also see like what service meshes are running in the cluster, and you can also select them based on name spaces.

D

So there was a decent ask like why we just get why we just show service messages. Why don't we show all of the workloads which are there in a cluster because, of course, measuring does measure does um like keep a track of which or um resources are there in a cluster? Whatever works will be run like deployment services, all that stuff it's being stored in the database, just that we never showed them on the ui. So um I was working on a dashboard. I was working on improving the dashboard here right now.

D

There is a new new card which shows all the resources which are there in your cluster current cluster connected cluster, and it gives you the kinds of resources and the number of them in your current cluster. So this I'm currently like it's almost done. um I just need to um need to fix one thing which, like I'm, not able to um select them based on the context selector apart from them. It's like done from my site. I would be completing it in like today and tomorrow. So this is looking like now.

D

You can see all the resources in your cluster then, and the number of them and the number of each time so.

B

That's the thing.

C

Feedback or questions for pranav, maybe suggestions on what might come next or like. Why are there 164 167 configmaps? What your config map crazy?

C

Well, I mean that's not pronounced, but just that's interesting.

D

Yeah we can tell you also like like this: is I'm running a watch command on my pods? So you can see. There are 14 ports right now and the exact same number is over here as well, so it works similarly for services, um I'm having nine services and same on my which are returned by my cuba.

D

Though, if you have more suggestions, please do throw them on slack or on this pr, so we can put them on some consideration.

A

Yeah one thing uh I mean it's picking, but uh you know the resource and the number like the number should be count since it says the.

D

And uh just one more thing I wanted to ask: should we add the service mesh table above the workloads, because the workload is too long and uh the new user might not know that? There's a service mesh people.

D

Service mesh list can also go long. I mean it's long for me, so I guess this is fine.

D

And yeah, according to this question, it's multi-cluster compatible like it will show all the resources based on what selection we make from here. So it passes context id to the subscription and then based on that it queries the database.

C

Any other feedback from others do you think.

C

Different visualizations might be helpful. Do we think that the table is a little long? Is there too much white space between the workloads heading and the start of the resources? Can you tell the difference between a resource and the fact that the column header looks the same?

C

um Do you like the centering and not left alignment in the first column? um Do you like the? Can you sort the columns the last stateful set? It has an underline underneath the uh even though it's the last item it uh it's I was just trying to. I was just trying to I'm just trying to pick out like something that's super silly. Just to to the point of like oh good, did you want feedback? Let's just hammer you with feedback.

C

This is great. This is such an improvement and um yeah. There's a couple, a couple of things to consider some stuff that might be a little hard which is gosh what? If what if people want to see everything or what, if they only wanted to see, select items, you know you can try to build an opinion in there.

C

Ultimately, though, you want to let people get to those details, um if should there be what what happens when people have a like, you know three or four more resource types: should we just go ahead and like set a maximum height to the table and make them scroll at some point that or let them kind of do that, but let them expand it or I know a lot of things to kind of play with, um but of what you've done and the speed by which you created this in a day which is great.

A

Just one feedback that, as we already mentioned about the namespace selector there, there would be namespace code resources and there would be non-namespace non-namespace code resources, so for them we can have inside of this card. Only the one so just to you know, have a uniform user experience. So if you scroll down to the service mesh part, you'll see that electron yeah here yeah this one. So uh we can have a similar namespace selector on the workloads as well, but the resources which are not name space themselves like the namespace resource itself.

A

They can lie outside of so there would be a card inside of a card. Maybe, and the inside card will be like name speed. You can like the service mesh card. Where you can. You know, click on multiple names, pieces and the different stuff would come up, and then uh there would be a static, uh non-name space resources which would be on the parent card.

B

Yeah look into that.

A

Okay, yeah thanks for the feedbacks everyone. uh Let's move on to the next item, uh minimizing the number of piles in the root director of machine case.

A

So this is here right.

A

In case he's not uh I mean.

C

Yeah mr yeah christopher's on christopher.

B

B

um So I'll go ahead and share my uh the changes that I made so my appearance, basically on minimizing the number of files and.

A

In the root directory of machine learning,.

B

Yes, in the directory of measuring, so um there are a couple of changes that were made, and definitely this might need a bit of I mean, may need a lot of review and yeah. Basically.

A

Okay, so you have uh created modules and some models to.

B

Compact, what basically happened was base. I was taking all the all the merchandise api code and putting it under one directory, so yeah.

B

A

Nice, do we have any feedback here? uh Anyone I mean it's quite a long. Video thanks. Chris. Are there much conflicts right now.

E

A

Time a pr would be any time a new pr would be much. There would be a much conflict with you'll experience some conflict here, so uh I fixed all the magic okay and have you run all the tests and build like machine ui provider? Ui build machine server and uh run a bunch of so you have you done your the end-to-end testing from your site?

A

uh Yes, I did. Oh that sounds nice uh um after the call I'll check out your pr and do tests from my mind and then I'll approve the vr. If everything.

B

A

Great, uh do you have any other comments.

C

I'll just say that I'm impressed with christopher like didn't. I wouldn't it's not a small change like the um like, and this might be christopher's first pr which is just like makes it all the more impressive.

C

um It was something that I was scared to do, because you get all the moving parts and- and you don't know what's referring to what- and I think last time that I checked out this pr and built it um uh now. I think I know it was successful like I was running measuring and I didn't see issues so I'm excited about this. This is great.

B

A

Great great day, thanks this uh okay, now now we can move on to the next item: uh user friendly api on capability registry.

B

A

So uh so, just to give a little bit of context in measuring we have uh we have what we call components. These components are uh basically uh okay, uh I'm trying to figure out at what level I should experience so I'll, explain it pretty uh high level. So we have these components. um These are there are too much so they won't be formatted properly, and these are basically a bunch of json, schemas and definitions that that we use and uh so like.

A

We have kubernetes components and each service mesh has its own components, and these components are a way for uh the adapter, like we have machine, adapters or kubernetes to let us deploy those things like. If you want to deploy a part through mesh to you. We need a pod component which we extract from kubernetes.

A

If you want to deploy an sdo mesh, then you need to have an sd, so these components can be found in uh each of the adapters directories and templates on workloads and at the time of uh when the adapter starts, it registers its workloads with meshi server and in mesh map designer and in other parts of meshi as well. We require these components so right now we have like only two ways to do it.

A

So there is this api home workload which throws uh everything that is inside of our registry, which is uh a lot of things or you can uh filter by name. So there is service dot code, so you will get that, but again, that's not pretty user friendly. So the idea was to let me open the pr as well, so I just created a thin abstraction over uh our current uh our current registry to expose some clean api, so just like uh kubernetes uses group version kind.

A

I have here use type version name, and here I shouldn't use type. I would say I was confused between registrant and type. So for now I have went with type so the apis look like this, so we have a slash api, slash components, slash, let's say um so. If we want to know what are all the resistance or types available right now, I have kubernetes theo core. If I pick any one of them, let's say sto, then it will return all these two components.

A

So let's say that I want to know what are the versions available for stu, I can say slash versions and it will return all the versions. Let's say I want to see what are all the components for latest version. I can say slash component slash, history, slash latest. It will return me this if you, if I want to say virtual service for the latest version.

A

So basically the whole thing is pretty much very much readable and yeah a much fine-grained experience in what we had before so slash, apis components last type, fit on that and then um yeah you can filter on type and then you can go all the way to version and uh all the way to name, and you can do it with um like. We have like our core components.

A

So if you, if you just pass like core or istio or kubernetes, it will return all of the components so and um if you pick, you can pick a specific version and you can use that.

A

Let's say that I want to see kubernetes 1.24.1 pod, so this would be the query and uh so basically a lot of filtering that was happening at the front and it's now you can directly. You can do it on back end now.

A

uh So if any, does anyone have any doubts regarding this.

C

Yeah the version seems potentially ill-placed, but um something that settle in, maybe in terms of the endpoint structure. It might make sense to put components under something like models, just because um components could mean measuring components and the system itself um things things to think on.

E

B

Lee, where do you reckon the.

E

Version should be.

C

Well, potentially, nowhere.

C

Like uh um yeah, I mean my my suggestion is that you don't include it in the endpoints that if you want to filter on that great otherwise, you just by default return the latest the components for the latest version.

C

It might require a little more thought. um It doesn't feel it doesn't feel right to me. I mean that doesn't mean we don't do it. I'm just I'm just saying.

C

It would be potentially a parameter potentially something in the um it's. It's a sub. It's a sub thing. It's like hey. What am I interested in um kubernetes um deployments, uh but only those components for this, maybe for multiple versions. In this version.

A

Right but it's pretty standard practice like in uh kubernetes, you have the four dot something slash v, one alpha one slash v and beta one that they their version, is it's just r version. It's not like alpha one beta one. It's just one point two or it's a version kind of version, so yeah.

C

It looks like yeah, it's a version, it's a version of their api, which we don't have a version of our api. So it's actually could be confused between the two.

F

C

Right would I I wouldn't necessarily yeah and as a user, I would assume that that would be the version of the measuring api since there isn't another version listed. I don't know it's something to think through it's something. To give a little more thought too.

A

That's why I put version after the name of the uh registrant, so kubernetes slash version would mean that this version is kubernetes. The core slash version would mean that our version.

A

Yeah yeah, so we will be moving this conversation uh in some like one stack and uh for later. uh So uh now we can move on to the next item, uh which is updates on the card view of applications and design or if you are there,.

D

Yeah, I'm I'm sharing my screen.

D

Yeah so um recently I've created a pr about configurations, ui announcement, so configuration involves application, design and filter. So this pr adds a source, type icon and a functionality in cardview also and fix all of the ui and ux bugs with code restructuring for all the configuration components. So let me demo this so previously we were just having the source type column in the tables. We were not having any kind of um ui ui show of source type, so we have this functionality.

D

So if we click to this component icon, so it will just download the token for with the source content.

B

D

Yeah I've also added a tooltip over here, so user would be like notified that this action can be performed on clicking on this particular icon and rest of the things involves: bug fix with empty state and code changes and with color enhancement of all this undeploy and icon stuff.

D

Yep, that was a quick update, so.

A

uh You are saying something.

D

Yeah, like I was saying like we have, I think, uxs on the call. So if anyone has any suggestion more great ideas, how we can um show more items or on the card, they are welcome to do and jump in figma.

A

Yeah call for ux designer sr, ui designer or both yeah.

D

So any feedback or suggestion.

C

Now luza had a good question on. um I don't know if it was applications or it might have been on designs. um So just a related consideration, especially around ux, so uh or users of meshary that are looking at designs.

C

They'll, see a very similar looking list of well of cards in either grid format or a table format. They can perform certain actions on the designs. So in the grid format you can deploy on deploy or edit, which are the same three actions in the table column.

C

um Here you can create a design imported design and those are top level buttons that you can do, whether you're in the grid, view or table view shortly, um root, raj and uh and four other people from aditya and uzair, and I can't remember, are doing uh or adding new capability to the designs section and to the filters section and that is to fork or clone a design, and so you can imagine there being a fourth action in the table column and that would you know, most of the time probably fit inside of someone's window their screen.

C

If you're in the grid view, though, and you go to add a fourth action- well, it gets a little tight and so there's a great open question about you know where, where to or how to update the design. This grid, designed to account for a fourth action.

B

Second, yeah.

A

In my opinion,.

C

uh Yeah, we don't necessarily have to figure it out now. I just wanted to raise it up as feedback, but yeah. If you want to go ahead, obvi and then I would move on.

A

Yeah so in my opinion, uh since we had the source type as well as download content available on the right side of the header of each card, so we can have that fork. Item, since we are doing something similar uh like fork, is also kind of uh sharing the design to someone else or just like downloading it and then sharing it. So yeah.

B

C

Oh I mean yeah. I missed that you actually weren't trying to answer that particular question. You were, you were calling out potential, meaning behind fork and the behavior of the action.

A

Yeah so uh yeah something similar, not exactly okay, you know, can you.

C

A

Oh okay, yeah should I say it or.

C

Either way, were you looking for anyone to provide feedback? Just then, are you just making it.

A

Yeah, so anyone here, please, don't write a feedback. If you have.

D

Yeah so in designs, you'll be having visibility and um um both in cards yourself right. Visibility should also be there.

A

Yeah, maybe uh start with object. My designs, I think it's green yeah.

B

D

I have shared a design uh for these patterns uh in chat, so anyone is welcome to jump and provide suggestions.

D

Back to thank you.

B

A

Okay, nice: uh let's move on to the next item uh updates on machine catalog capability.

E

B

E

E

Yeah, so uh so, as as we already mentioned, the uh the goal of this feature is to be able to uh clone a um so. There is still a discussion on the uh the world that we would be used, but basically clone the patterns and filters uh to your own uh account. Let's say if you have machine cloud, then clicking on fork on someone else's pattern would make a copy on your account, so uh it will create your personal copy. So this is just a demo.

D

E

Part has been worked on by user and aditya, so I just bought together some ui just to show the functionality. um So let's say this is this is a pattern which is uh this. Is I own this, but it's a private pattern, so I can make edits to it. If I want and I can save it and it saves just fine- and this is not my- uh I don't know this pattern, so I've also added uh authorization checks. So I, if I, if I change this, it will not let me it will not.

E

Let me uh edit the pattern uh and also for now, I've just replaced the delete functionality with fork and it's right now. You know it's such that you can only four public patterns, so this one's uh this one's a private pattern uh so yeah this is a private button, so I won't be able to fork it by click fork. It'll say you know not serious to fork, and since this is a public pattern I can I can fork it. If I just click this, it will create a new copy, and this is now my pattern.

E

So I can make changes on to this and it'll yeah it'll go through just fine, so uh so what we want to do with this feature is to integrate it with machine cloud where we can have public patterns which is which are part of merchant cloud, and if a user wants to edit that pattern, they can first clone it and then they can make edits to it.

E

I just watched and changed the functionality.

E

The server side, I've added it as a capability on the machine cloud side, so it works as a capability. Now.

A

uh One more thing that just the snag bar error: that for one error I mean it makes complete sense, but but those errors are mostly like. You know something went wrong and this is a completely uh expected behavior if you're trying to edit basically and non a public thing so the on the ui side. If someone tries to do the edit this thing uh and you get this 401 error.

A

Basically, if would you get unauthorized error or instead of saying just showing the snag, but there can be a better error that describes that thing that this is public, and this is private for people who are just you know playing around with it and they get this 401 and instead of thinking that okay, so there is this private and public, uh we show them actually some actual information that some kind of an info like something like disclaimer info type thing a pop-up. It says. Actually you can't do these things.

A

Look for these labels, let's say public private and those are yours, and this is how to fork it- something some small text there, because we know how to do all this. These things, a new user, might not.

E

Yeah, so I think a better idea would just do not have a full button or a save button in case where it's just not, for you know, authorized on your site.

A

But how does that.

E

A

um Yeah, that sounds good, but uh on the front end, do we know beforehand that this pattern is uh yes, okay, then good.

E

Yeah, I think, if there is anything to share anything, to share.

D

So only thing for the uf was remaining is uh adding catalog feature in cards view. So today I added that as well and most of the parts for the ui is done only if the rigorous pr is. If, if uh appear, is much, then we can perform the final check and merge this ui changes as well, and then it will be completed, and there is only one question from me like in, like, as we are not allowing public designs to be updated.

D

uh Is there a check added to not delete the public uh public catalog as well? Did you added that in cloud or not? I don't think so.

E

uh Okay, yeah. I think I missed the delete part today.

E

I'll add this right now: just wait: okay, I'll check in there.

D

This so when the both the piazza merged, then we will perform the final check in the ui and it will be completed.

B

A

Great uh so, uh let's move on to yash meshvery, structuring branch updates, open tears and trackers.

D

Yeah, let me share my screen.

D

So um a lot of work has been done on the restructuring part. uh Most of them are merged or open prs or are in progress right now. One of the main thing which needs some focus is the uh is the backend part. Mostly the graph will uh get get queries because we don't have uh queries for a pattern.

D

uh I think filters and applications as of now so, and we also have to update the whole uh restructuring branch uh and back I I tried copy pasting, the whole backend and I think it worked, but uh there was many great issues, so uh I would need a little help with that.

A

B

A

You have the photo, I mean the restructuring branch uh live.

D

Yeah, uh I I I have it so um so for patterns and performance. I've made the queries and it's stretching data from the graphql.

A

D

A

I think we can only see the uh the excel sheet.

A

Are you showing something else.

D

B

D

Now you can see right? Yes, yes, yes, yeah, so I've added, graphql queries for our patterns and profiles. I've also added for applications, but it is not connected as uh we don't have a get query endpoint for it.

B

A

Okay, uh good so far, so uh do we have any feedback here.

F

D

Like we have the subscriptions for patterns application of filters in mesh map, so we can move that here as well. Initially right, we don't need to add new grass, basically.

B

A

Yeah, so I I I for for these pages like uh are you using graphql subscription or uh uh are you choosing rest api endpoints? Oh.

D

I'm using graphql.

D

So um yeah I'm I'm using draft already to get to the patterns.

D

And I think, uh update from others, uh I think uh leonard is not present. He has made the performance page, the uh dashboard uh he hasn't made a pr as of now he just shared my screenshots and updates from akira and shukrati. Bikers I will let them share.

A

Okay, uh by the way uh uh this one, uh you know the thing we just talked about: the open, vr and tracker and all so. uh These are like very open to uh newcomers and everyone who wants to try out uh machary and uh want to contribute to machine ui. So uh very like very soon we are moving to the restructuring branch as our main branch, and we will be using that uh since we are on mui b5 on those branches, so uh yeah.

A

So if, if you are like uh interested so yash is a great man to speak to so basically, I'm calling out for volunteers if you are uh interested in ui and react and mystery and want to contribute there.

B

A

Good nice, uh you can move on to the next item: uh creation of reusable cards for our ui restructuring, pr6051.

B

Right here, yeah, I'm sharing my screen, so uh so I've created the this divisible card for, uh like this card, has been used for like uh application pattern and filters. So I made this part irritable and the another update from my side is.

B

uh I was working on this like in the docker extension like uh like telling the user like which mesh adapters are notepad and it's sometimes sd or like named or it's not working. We don't tell the user about which mesh adapters are not present currently, so I was working on this vr. uh Actually lee pointed me like. uh Firstly, we can use something uh the rest api endpoint for that, but, uh like that kind of first law, but I used something with docker.

B

The docker provided the sdk tool like uh so what I did like here. I just dish the container and filter it out and all that stuff now, so, where is that? So now it's like working so quick like you, don't need any any like feature or anything between between it. So only thing that I am facing difficulties like uh if the user hasn't like uh enabled this, so it won't work that only thing that I'm facing difficulty now.

B

Any thoughts are prepared.

C

I missed um the quite you just like you're, going to use the um rest api to determine whether or not these are.

B

No, I'm not using the rest api, I'm using the the second approach that you've provided me like you use something that is provided by doctor docker uh doctor like doctor. What do you say like uh means like.

C

This makes sense.

B

Yeah so now it's like working so quick, like uh it's gonna, make the user experience like much more great.

B

So only all the difficulty I was facing, like the user, should have to be enabled this auction, or this won't work. The docker command won't work.

C

Okay, that's not a good solution, then that won't work.

B

C

um I wonder if you're using the yeah.

A

uh Probably calling out on uh the docker extensions engineering, so it may.

B

A

From their cycle or if not.

B

Yeah we can change the solution.

B

A

uh Spec for machine app import command.

G

Yeah I'll share my.

G

Okay uh is my screen visible now? Yes, it is uh so basically, uh last couple of days, I went through this spec for the misery, serial app import command and I'll be starting on the implementation of this. So before actually implementing. I want to discuss how the command will look like like what will be the flags and uh what will they do and uh and yeah so so, like you know, there is a command called mystery, ctl app on board.

G

So what it does is basically, uh when you provide a application uh ym into it, so it imports the application and also deploys it. uh So this app import command will only be a subpart, will kind of be as a part of the onboard command and we used to only import uh the application. It will not be deploying the application.

G

So uh here is how like it will look like so we thought of mainly three uh flags. So first one is file uh in the file command.

G

You can put the path to the your file or you can also put a url um to get a url like similar to the app on boot command, and then there is a name plug in which you can specify the name of the application like in the app on boot command it's uh by default, set to the file name, but in this command like you'll, be able to modify the name and there's a source type, so you can specify any source type that is supported like like a kubernetes, manifest or a docker composer or film chat and and yeah, and it will be used to import the application automatically and also along with that, I want to discuss that we'll be making some changes to uh app onboard command, also as a part of this, mainly to make it more consistent with this import command so that they have similar set of flags.

G

um So mainly, I suggested three changes, so the first one is that we should add the name command to on board as well like currently uh like. There is no way to change the name of the application in the onboard command, so we can add the name flag to that command as well, because I think it should be similar like if it's supported in import and we should be also supporting on good command and uh and then the message and we can use messages with the application name to deploy an already imported application.

G

So it's this functionality is already present, uh so nothing needs to be changed, uh but we can improve one thing here that uh so, let's say if a user types in mrcdl app onboard with app name and also provides a file with with f lab, so it it ignores that f lag and just deploys the application with the happening. So we can throw an error there saying that, like it's ambiguous, we don't exactly know what user is trying to do.

G

Yeah. That was all.

A

We are good uh additive early or anyone else on the call any feedback regarding them actually secure.

A

Okay, so he has a feedback about what he. What we really need is a thing like uh this machine console: okay, he's referring to some git command, so yeah.

C

Well, yeah well, or is to say, like hey: we just have some like a horrific ux going on on a lot of the commands, and now harker rat has um an open pr, um alpha and nishant, and a few others have jumped in to try to help make sure that we have a decent response.

C

C

Do it. There's a we've been talking about it for a long time. um Just go. Go, get like that. Just should never be. The response. Error requires at least one args. I am a robot, don't use me, I don't know what to do like it's just terrible and so there's any number of better answers.

C

And then there's even better answers, which is probably we shouldn't, give people a link. I know it was originally my idea to give people a link directly to the specific doc doc set for the documentation page in mesri.

C

um Well, docs.net.io.

C

The thing is: is that was before we moved this source, the single source of truth for the docs right into the cli itself, so other than a screenshot there's, nothing more that the user should get from the docs. There's no reason to give them the link. It's just like you actually want it to be pretty concise um in terms of what you show back.

C

So that's one one item um of what nashant was saying is that you do you use the word consistency or consistent um a couple of times? That's exactly right! You can hardly go wrong with making it consistent.

C

In fact, the commands are not consistent, which is causing, because it's kind of a big deal like if you're going to use something like measure ctl app on board and give an app name great, just be prepared to lean heavily into measuring ctl perf and how it is that it does search for a profile name, um because you can have all the same challenges. What, if you have two two applications named the same thing or um or you can't find one or like there's a manner, a whole series of interactions.

C

You would have with the user how you display back multiple results, and that kind of thing, which is all great and a lot of that is handled for inside of mesh ctl perf, so copy that ux and be consistent.

G

Actually, that logic is already present in onboard like if there are duplicate app names like like the user, is shown an option to select an activity.

C

That's great yeah, that's great.

G

Good and also like about the links I want to ask like like instead of links uh like yeah, like you, told that we don't need to send users to the an online link, basically because we are generating the documentation from the cli uh itself. So uh can we like, instead of giving the links, we can uh say the user to use the help commander like uh so? Let's say there is an error in messages system something so uh like.

G

Currently, we are saying the user to go to docs dot message, dot, io, flash message, ctl slash reference, slash system, so, instead of that uh we can say the user to just uh like one more information like for more help. You can type in messages here system help like uh yeah.

B

C

Yeah, I'm good the stuff that you said has merit it'll, probably take a little bit of a longer conversation, because the first part was correct. The second part, I don't think so um we should there is a um ux design.

C

um I don't know if it's a guy, a style guide that we've been trying to publish for measuring cli, and uh it's here I think I pointed out a time or two. um Let's refer to the style guide. With respect to that particular question and let's um um update the style guide if needed, as we go to figure that out, because that's a pervasive question: it's a really helpful to get it figured out and kind of made.

B

Uniform okay, yeah.

A

So the next item in the list is uh components on machine docking, compatibility, metrics and it's different.

F

Yeah I'll share my screen.

B

F

So, as I actually spent earlier, uh the components are a bunch of json schema that has adapters to. You know, uh help us to add some information on the adapters. So what currently I'm working on is to consolidate all this information about different components as components of different versions like this machine linkedin has different versions and inside them we have different number of components. So what I'm working currently on is to consolidate all this into a single single page. You know measuring docs, so so to do that.

F

uh Currently, we I have implemented a workload uh uh action workflow that you know copies all these files in present in all the adapter repository and transports them to the machine, docs folder in measure repository so yeah currently, so this is a to so display this table. I have you, know kind of implemented this design and I'm looking for suggestions on what else could we make better out of this like if it can be made better?

F

So I have been made a similar version, similar uh design as a compatibility matrix in which we show all these details, and if the user clicks on one of the rows, uh he is shown with all the different components of this particular version, the latest version, and if the user wants to see the integrations and components of all the other versions, he goes to. He clicks this button and goes to another page so yeah. I want any like any suggestions or feedbacks. If there's something missing here, that would be great.

C

So yeah so does divine and alexis and others who are um well. You know anyone, but but specifically those gentlemen or others who are your user interface inclined. We have an information architecture challenge in which we've got in a good way. We've got tons of information, we need to display to the user, but we need to do. We need to make it palatable and understandable. We need to summarize and allow them to drill in, for the most part, that's happening already, um but betty. If you go to the I might be jumping the gun.

C

I I guess I expected you were talking about the compatibility matrix and the integration test. Did you already talk about that.

F

C

Okay, that's great so then, hopefully divine and alexis will have their ears pricked and listen as you go through. That.

F

uh All right, so I guess I'll keep on working on this and make up yeah. Then we can have further suggestions or configurations in that viewer itself.

C

Where will the see all versions go, uh sorry which one the hyperlink click to see all versions? Oh this.

F

This will show all these uh so currently the we in this first page. We only saw the latest version of particular adapters, so once we click this, we will see all these all these versions. In one page, all these version and their integrations in one piece.

B

C

Yeah, oh yeah. Hopefully there isn't that performance hit on a jackal site or something like not on not for users, but I mean in terms of building and stuff like that, like that, that's really that's really fantastic. um It.

C

It might be overkill, like maybe hold off on it until a couple of users start saying hey. Why aren't you you know? I can't see this for the older versions. I gotta, um because this is fairly impressive as it is the like. The documentation is up to it's pretty darn detailed per adapter per area.

C

I don't know I mean just before you put in that much effort. It might be good to get okay.

A

Yeah, oh so this thing uh we created this out of, uh I think the only pr that got merged was the linker d1, so the sql osm data is.

F

Yeah, it's uh statically, I just created for a design. Okay,.

F

Okay, so moving on to the another topic that I have, I was talking about the compatibility matrix. So a brief overview of what comparative metric is that we run several e to b tests on different adapters when, when any changes are merged into these adapters and all the test results are, you know, consolidated and reported back to the mystery dogs uh like if they, if it's failure, we show it red if it's breathing green.

F

So all these tests are, you know, uh stored in one piece in mexico, uh so we can see all the patch results also. So from these all these integration tests we we have, we have this competitive compatibility matrix for different kubernetes version. So currently these uh the comparative matrix is kind of hidden right. It's not uh shown directly to the user in the menu or menu or uh what I'm trying to say is it's not currently shown to the user. Visibly you usually have to go to different level.

F

Steps to you know see comparative matrix, so what we would like to you know what we would like to have is that show these compatible metrics in an uh in the in a way such that user can access it easily. So you know kind of upliftment of comparability matrix.

F

So we are looking for any ideas or you know, suggestions on how we can do that, how we can easily incorporate the that thing into the mastery doc so that users can easily easily be compatibility metric, so any suggestions or any design ideas that how we can incorporate that uh that would be really helpful. So since we have many uxers and ui in this call, so that would be really helpful. If anyone has any suggestions about it,.

C

Yeah, I said divine: does this make sense? Is this something you think you might want to work on.

B

uh Yeah, but I'm still, this is confused. um Where are they supposed to access the complexity methods from exactly you said, the probably let you go. Oh.

B

F

B

Cut in between, I keep on clear.

B

F

Okay, so so you wanted to know how we can access compatibility matrix right now: yeah yeah, okay, so if you want to see compatible matrix, we uh we can do it with the url talk.machine.tile or you can go to supported platforms and then click on this compatibility matrix link.

F

So we want to know move away from that. You know and show visually or so readily combine the matrix easily. I mean possibly in this page itself right. This page is uh one of the option to show the compatibles.

F

uh Sure yeah that ping me up in slack or in public channel and really happy to collaborate.

F

And yeah, okay and one more thing I wanted to point out like uh there are lots of reds in the integration test. That is not looking good. So if we, I guess, that's a failing, I think we have to look into that. Also, uh this thing is uh sorting issue that will picks up here as much, but others are so important. Yeah. That's what I've already pointed out.

A

Okay, thanks yeah, so someone needs to look into the feeling. That's why they are failing on each screens.

A

Okay, so I think we have pretty much covered all the topics today, uh yeah great meeting so far. uh Another announcement that uh we have to make is uh the lfx internships and around seven of them is open uh for you to apply. You all are like uh encouraged to apply to lfx internships, and uh you know show work and.

A

Fill the application form enough so yeah we are waiting for your application, that's what I wanted to say and uh yeah. I think we are on the top of the r. So thanks everyone for joining once again, and we can conclude it uh here so uh lee do we want to uh say something.

C

Oh yeah, uh with respect to the internship, the lfx like yeah, hey, it's the it's like eight eight fold over, I think the most popular lfx internship, um so yeah go, go apply. It there's like seven of them out there.

C

It's something that we do all the time. There's any number of individuals here that have been through either g sock or g, sod or lfx, or just a direct internship with layer. Five um talk to them: either read their blog posts. A lot of them have written blog posts, not all of them.

C

uh If there was someone that had a poor experience, I would definitely point them out to you. There are. There are a couple of people who have started internships and we just um didn't position them for success, and so they had they struggled.

C

Come see me and I'll give you their names, you can go, ask him how it was.

B

Yeah great interrupt, I think, a little bit.

A

uh Actually, this lfx internship- I actually had filled the form a couple of months ago, but uh am I supposed to give the application.

B

Or the form to.

A

B

A

B

Of the mentors as well on slack or something or I just need to fill the form on the site, the tasks which are given to me.

C

Yeah, what you know what the inside say? Actually I should go, write something and go po. The way to get an internship is to well go talk to uh the guy that just presented or abi or ashish or there's like 15 people on here been through internships or in one now they will tell you. Internships are earned not awarded like come and enjoy the project and get involved learn a bunch of stuff, as a matter of fact, whether you get an internship or not you'll.

C

Take the same experience away, there's just like a lot to say on this subject but like if you want it, come get it. I'm not. I'm not kidding at all, there's so much more to do than just an internship by the way. Just go go talk to the other folks, so anyway, I totally. I would love for everyone to be interning right now or, like you know, everybody to own a piece of the project like make it yours.

C

um The internships are just to like support those people in owning those areas. We've got actually. We've got mr mesheri cloud who just who just dropped a chat message like pranav is one of uh is like the main guy behind that component. There are other people who contribute and engage, but I'm just I'm pointing out him out as an example of an individual that came and wouldn't stop asking questions, and they wouldn't stop like making good contributions, and it was just like that's perfect.

C

We just want to support that, and, and so um yeah that's just about as concise as I can be. I think, on the I never look at applications for my part, um not once I could give two shits, it didn't tell me anything about what you're most of it's a is a lie anyway, which really pisses me off. I gotta tell you like: that's one thing: don't lie uh anyway: okay, good, so I'll, just prattle on for like a long time.

B

Yeah so yeah the.

A

Crux is like uh uh join the slay slack. Do some contribution uh get involved like the project? If you don't like that, it's fine, if you like yeah you're you're in for the internship and even more so great, uh we'll carry the rest of the items in the next meeting uh till then uh good day. Bye.

D

Bye, everyone I'll.

B

See you guys bye.

C