DASH Community Weekly, 2 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: DASH Workgroup Community Meeting Nov 2 2022

Description

Volodymyr Mytnyk presenting DASH test plan (Intel contribution)
PR217 - Added Overlay Test Plan

IPv6 ACL - environment related and resolved (one missing line in the pipeline)

Reshma: https://github.com/reshmaintel/SAI/blob/dash-ptf/doc/SAI-Proposal-SAI-PTF.md

DASH changes for SAI PTF - OpenComputeProject/SAI repo
PR #1573 – SaiHelper changes
PR #1592 – Test cases updates (Underlay)
PR #1590 - Add DASH API

Reshma: Work w/Richard and Guohan to expand the testbed to support the DASH testbed

A

Vladimir and Russia take over.

B

Everyone uh voldemortnick will be presenting the test plan for dash. We had a few discussions earlier where the intention was to have something of a plan uh test plan overall for Dash and uh and the suggestion was to use what we have or been working on as part of that test, Hunters that we were building and the test cases. We also had this test plan that Voldemort meeting. Would you like to preserve.

C

C

Yeah, so do you see my screen.

A

C

Okay, so today we're going to discuss a little bit, maybe just give it a brief overview about this pull request, so we created, like probably a few weeks ago, I think uh and on actually mainly work in this, and we created the test plan for ETF test cases uh to cover the let's say for starting from the v-net use case. So there was a couple of comments already. Some of them are undressed, but I want just to go through through the pull request through the changes.

C

What we have there, so, let's maybe switch to the view okay. So this is the folder for the test plan and actually the test is the folder is a root folder for all these cases. Also, all these cases that we are adding uh for the dash PDF test cases. We also cover them in the test plan. So it means that anyone else who is adding the test case or updated so should update this test plan accordingly and just you know, to keep in sync everything here, so we have some.

C

We splitted this test plan for overlay, eni connection track in the net tubinget and we can like go to each of them and to see what are there. There is some requirements: performance stuff which is just a to-do stuff and there's a test tubes that describes what we plan to to test and there and if we go, for example, dni we separated it out.

C

We also have some description, there's a brief description of the test cases and we have the list of use cases that are covered, and we also specify here what test cases are related to this particular use case. So you can find this name in the at this case. Script itself.

C

Okay, and there is some brief description of each of them- we're still working on this pull request anyway, so we are updating and at time to time. So uh if someone is interesting in this, I mean updating. uh I know that there is some work is going on with alcohol, so this test plan also can be used for, say, updating and powering the additional alcohol use cases and yeah and Venus to the Nets. uh I'll show it.

C

So what we have here uh test shoots, album, throating and, uh let's say test name in the scripts and a brief description, and the daily description is over here in bed wrote In. This case is some integration tests, negative tests scaling still it. It seems to be defined and some questions that I needed to be clarified before moving moving.

C

So uh this test plan is actually in review at this moment and what we ask Community to to review it, because we would like to have it merged as soon as possible. We understand that this is still like it's a beginning right. We will be updating anyway, but we want to collaborate on this. For example, if someone trying to add new test cases so would be really great to have the test plans also updated and collaborate on the theme, and briefly, that's it from my site. Refreshment. If you add something.

B

Yeah I think um there is a lot of content here. Please let us know if we should go through uh test plan sort of a review in maybe uh this type of a call or a submitting subgroup. You know during behavior model or something where we can walk through each step and collectively decide and add more or please review it offline, and then maybe you know soon as we merge this uh merge. The initial version that we have here we can add more to it.

B

I think IPv6 Apple was something also that was newly being added by by Hanish, so yeah. We can just add all of that here on once we merge this and create new pairs on top.

B

A

I have no objection, it looks good to me.

A

Chris I know you joined late. um What do you think? Yes, these are all under the test: folders um structure, yeah.

D

I think I I think I reviewed it and okayed it if I'm not going to take them. Okay, cool and uh you know- we've been working on test cases as well and there's a lot of alignment. So it's a good thing.

B

C

Good common crease already so crease somewhere Above This pull request. Yes,.

E

It will say fixed I.

D

Fixed I'd probably need to go in and uh approve, then yeah yeah for playing ketchup with each other.

B

And looks like Prince is also looking at this one.

C

F

I'm I'm going to review this yeah.

C

Okay, so if there is no any other objections, so I will just stop sharing.

A

G

A

The congregation yeah.

B

um The ipvc cycle uh there was an issue raised by um honey earlier I think um Voldemort and Anton went through that and it it turned out that there was no change required in the ttf harness, but there was still some issue that had to be fixed, um yeah and that's result. Now: it's environment related yeah.

E

Yeah, thank you, yeah already merged, even so that sweet this we discussed in the mail, and uh so there was a solution actually in the test code. So.

D

Yeah one line, one missing line in the python.

D

That's that's. That's yeah, we're past that.

A

Okay, uh anyone else have a topic for today or PR. They want to look at.

E

I have a topic except of PR oud, okay yeah. So that's the question to maybe to the prince so that everybody else so maybe there is some updates on the Sonic management part, maybe some uh visibility when we can have some, maybe over latest plan, to look at it for the Sonic management or some updates to the topology.

E

Do you have any.

F

Sorry, can you repeat the question yeah.

E

The question whether we have any updates Dash related updates for the Sonic management test framework, maybe some the most interesting time like I'm, most interested in uh our latest plan. Yeah.

F

Yeah I think I think the the peer should be ready like this week or early next week, so the the team is working on uh on creating the Sonic management plan. I think that for the test plan, um I believe we are referring to the dash a v-net doing a test plan that that's available today.

F

So there's no separate test plan, but the implementation of the of the Sonic management will be ready by by this week or next week.

E

Oh okay, nice yeah. Thank you yeah. Thank you very much.

A

Great, thank you so much anyone else before we go.

A

I, don't see anyone new on the call for introductions. I I mean I, didn't introduce Kishore last time. Kishore. Do you want to introduce yourself since you're new.

H

Hi sorry I was muted yeah. My.

G

H

Kishore from Marvel I think I know many of you through various uh other engages that we've been having our past glad to be here. Part of this project.

D

Hey Kishore good to see you here, it's Chris! Thank you, sir. Hey.

H

A

Then I'll stop the recording, I think that's all we have for today.

I

I I remember: there was some discussion emerging those tests into the Sony management right so where? Where is that that effort going.

B

um Go on this test plan that we presented, um oh the other, the test cases that we have been adding right: the winner to Win, It Etc,.

I

Hey you know, like the the you know, to set up the test pair right, so so where? Where is that efforts going?

I

So you know like running those those tests from the Sonic management right so.

B

I

There's not much efforts on that front, because I think you know it's hard to maintain these two different tests framework.

B

Right so um I think Prince had mentioned that the winner to win a test test cases that we have been adding that will be used utilized in the Sonic management. There is a threat going on about that uh review, going on as well.

I

No, what that means that you know those those currently this uh this attached attached. The side tests should be also be able to write in the Sonic management right, so you know set up the topology. Those kind of things.

A

I think our overall.

B

A

Bed infrastructure setup Etc: there are plans to dovetail it into Sonic management want like in about four months or so wasn't it Chris I'm, Richa, yeah.

D

I think that was the discussion that we we still need to get these tests. You know even working um on behavior model and you don't actually have you know dpus that can run into Sonic management test bid right, there's not even standard Sonic interfaces. Yet so it's a little premature.

A

Yeah, but it's it's definitely in the plans we just needed.

I

No no I I didn't mean that I mean you know even for those size shrift test right, so you need to set up the test. Bed I mean you know. Whatever the test that currently can be running, uh you know you know setting up those test pads that should be a line so that we can, you know, use a Sonic management to set up the test band run the test all right, so.

D

Okay, well, we don't even have a dpu. We can run tests on yet, but.

I

D

I

The current those Dash tests, you know well.

D

Publicly uh bmb2, but there's also some tests being run, let's say out of communities eye on vendors implementations, but those are not really available and for vetting right. There's no public way to vet any of the users. Those.

A

Are closed discussions right, yeah.

D

Yeah people are doing work and vendors may be implementing bits of this on their own, but it's not really public right. There's, nothing really, officially ready yet so.

I

We can't really.

D

I

No, but the test bed is set up. Those kind of setup can be aligned right, so I understand the implementation may not be available, but the you know the um you know for for the for the whole test setup that you know we should drop the comma framework to to set up and align that with the Sonic management right. So.

D

Has anyone made any progress on that that they can disclose in terms of having you know, implementation that runs in that framework.

B

Oh, the framework was uh tested on um Intel's, ipu, dpu um and then uh only uh upstreamed, so but yeah it is not public. Yet.

I

Yeah yeah, but no I I, don't think we we we. We need that right, so I'm just saying the framework how to set up, for example, whatever vendor uh you know we are working with or the you know, um anyone that, for example, if the vendor wants to run the test right defined by this, uh you know this stash test. They need a way to set up the test bed right.

I

So then, for those things, I think you know we previously agreed to align with the Sonic management to use the same way to set up the test band right as I remember. There was earlier discussion on that, and then um you know that that we get aligned so that people who ever familiar with the you know traditional Sonic touch better setup, can use the same way to set up the test band and run the test right. So.

E

So, as I understand, we need to have like something kind of ptf2 topology type, because we have PTF 32 PTF 62, but we should have small one exactly.

I

Exactly yeah yeah so.

E

From the commit uh that mentioned princess so that there was a new topology Appliance for two ports, but this is the another one, so ptf2 or maybe like Prince. Do you have something like that in the roadmap to have the aptf2? No.

F

And then PTF is not in my roadmap.

G

F

Right now, yeah focusing on the the actual test bed with fan out sander.

B

There is already a test bed with uh Richard right, I. Think Richard. You has a test bed working with Rita for the PTF. Do we want to augment that test bed? Is that what you are suggesting? Gohan.

I

Yeah and also you know, I remember this- uh this Dash test- he also uh you know, connect with some connect with this uh traffic generator right. So they you know. How can we Define that and you know, then people have a standard way to look okay. How do I set up the test but physically and then run set of a command to provision the test bed?

I

Then you can run the test right so I mean we can start with a PDF which is already available, but they you know, for those for those traffic generator based tests which the you know we can also. You know, move to the same test batch. So they can. We can write them, intend it right so.

E

I can't believe that, right now we can add only some topology files for the PDF tools. That's it because, anyway, we do not have anything like CBS for the Dash, so the current BMW to its not the same as IBS, so it might be like some future future tasks. Once we have some Hardware that we, then we will be able to prove.

E

Yeah because like so, you see that right now, if I want to try, for example, any kind of topology I'm able to use IBS for it. I can uh you know I'm, just gonna like topology on the VMS, but for the dash there's no such stuff. So like the only thing that we can do is like the same as Prince commit for appliance topology for the PTF to the body. But every vendor will need to like to go visit us on their own to set it up for the hard one.

I

So so this uh you know this side handle is Dash enhancement. Then what? What is that? What is the test bed set up there.

E

B

uh We have, um it is the same. uh You know we, we need um basically uh Two Hosts uh and um we have the client of the PDF running on one and then the server that has uh also the dpu in it, and we have uh mentioned uh the topology diagrams that are required in the PTF document. uh So far. uh This is how we have tested, and um instead of using a dpu in the second host, where the server runs, we can use BMV too and um yeah.

B

This is all that's needed from PTF side. uh We have tested it on the dpu that we have.

B

And and I think we have already merged uh all the code changes and the HL APR is also merged into the tie that just went out uh 1.1.10.

B

um That has all the dash enhancements in PDF.

D

Oh restaurants, that mean we can change the sub module branch in the dash repo from your fork to the main.

B

You could, but we will have again more and more changes here. So it's up to whatever hey.

C

Chris Chris, so that is one change for the PTF itself, which is chain changing the pdf version in the side report, so so range needs to be handled before the Main.

D

Right so the answer, the quick answer is no we're still waiting for that to propagate up to psi. Main.

C

uh Yeah I would say: I would expect yes, but.

I

How we started those discussions with Siamese, you know we should I mean if we are not expecting those the header changes. We should you know the I. You know have that discussion and get into merge into the main right. So right.

B

Right first, uh one thing is that we presented all the changes that we have done uh for for dash in PTF to make the entire framework itself work with the fewer ports and also how we were changing the tests and all those things we already presented in the PSI community. That was about a month ago, and then the hldpr was already merged to site, and the code PR also was merged into PSI 1.10. It's already there now.

I

Okay, well then, what what prevent us, the you.

C

I

C

The let me explain uh so uh PDF PDF framework is using the PDF to generate the traffic right yeah so, and it includes the PDF repo, some some old version of the PTF. So we are not talking about the PDF framework that we changed, so the changes for the PTF to run. The this case is on the let's say: Dash is already merged.

C

Okay, so the framework changed as I mean, but when I'm saying the PDF so I'm talking about the PDF as a traffic jam, which is included as a sub model in inside PayPal, yeah.

J

C

Yes, so there is a one change for the PTF as a traffic jam itself, which we found one issue there. So uh this is why we still using uh the the custom repo this moment because it includes the fix for the vx1 packets. So once that fix for the pixel packets are matched in the PDF itself, then we can move. Let's say PTF inside repo to some latest version.

C

D

We have a are we working on getting the P4 Lang to accept those changes into PTF? Basically,.

B

Yeah we can exactly what I was going to say that you know we can have internal discussion, and you know the before line co-chair is already here. We should be on the same page and I think uh yeah. We can take that further. Get that completed in the present it on the Friday before line meeting.

D

Right so we have, we have to propagate up through two sub-modules to eventually change what Dash is using yep.

D

We need P4, Lang, slash, PTF project to be fixed and then merged into Main, and then we need to import that into the PSI PTF uh or as a side test PTF sub module and then merge that into simane. Then then, we'll be able to use simane in dash, did I.

C

D

That correctly.

C

Yes, yes, Christopher, okay or we can. We can think about how we can you know, maybe change the PTF only maybe it's also an option the option, but what you said here this is the right person, yeah.

D

So that's kind of illustrates: Gohan, like we were talking about an earlier meeting. Sometimes we need to use branches until the the external process works its way through and that can take months.

B

Yeah this one, we will try to push on our password.

I

Is Dash test right so because I'm trying to understand the test that on this PR 259?

I

um So there are this PTF test size shift test, and there are this uh test the side we're not in a different test folder. So what was the difference between these two tests and the uh you know? I know the the PTF test can be right in the bmv2 environment right. So what about the other tests? How do we run them? Where do we run? What is the environment to run them.

E

Into environment, if I understand correct, you are about that. The one that we presented last week, so that's still being with the environment, still the same, so only that it's introduces Snappy traffic gen, uh that's that's it like, uh but the setup is the same. So still we have like a b, let's say dut to Ports. We still have for all other blips, but like uh replace.

I

But then are we running them in the uh in the CI part? You know in the CI integration you know anything for the PDF for the all the tasks that have been running on that on that right. So.

E

Not running because they're not merged.

I

No, no, no, oh, are you saying? No, no, this part I, don't get it I mean you, we, you know the environment even in the pr remember you know for the PTF uh they you know the pr can test it right. So all these PR can be tested tested. This test can test those PR. So why? But why? You know I think with their merch or not those tests. If, if it can be run, we should run it right so well. What is that? Maybe I didn't quite get this part.

I

Why why it's not even what's the environment to run those tests and the.

D

Are we talking about PR, 217, I'm, just I'm, getting a little bit lost, no.

I

Not too too, no, not 279 right so.

E

Oh um well maybe like that's another issue. You I mean that we have tests folder and we have Dash pipeline folder. Both contain test cases and.

I

Yeah yeah yeah, yeah, yeah, okay,.

E

Okay, oh okay, so let's say I will explain then so we at Intel committed to the to the tests, folder Dash. We need this case and yes, we expect that this will be a proper pass for all test cases in the future, but completely yeah. So, but we currently, we are not running them on CI, because bmv2 has some lag features.

E

So we already raised up this question on the bmv2 model meeting so that some underlay calls are not supported, even even if there was just even issues getting number of ports, which is uh extremely important for the PTF framework, because we are getting number of ports and then creating some objects in the PDF and so on and fbm V2 doesn't does not return this. This is like an issue and that's why we committed that this case to the proper pass, and we expect in future to add all new test cases there.

E

But right now the running of that test is disabled, because bmv2 Readiness so like once all blocking issue are fixed. Definitely we will uh have to move to the have to move to that new, proper location. The final one in the test, folder and we will commit even more test cases like and enable them on the CI. So at this stage yeah, so that's I opened a few issues uh regarding that, for the uh one was fixed if I'm not mistaken, but there are a few more.

I

To which test you said you, we cannot run the one on the test folder, although the one on the uh the dash pipeline test, folder.

E

So the display plan, so let's assume that this is kind of a test for environment, just to ensure that environment Works everything like is imported and so on, because the tests itself are written was the bmv2 only so that's on the hardware. The behavior will be different and the in the tests folder. We are we this cases that are, they will see Hardware oriented, but the problem is not because car like the problem is that bmv2 does not support some calls that that it should and yeah. So that's uh planned to fix.

I

So for but okay, but once bmv2 fixed my understanding. My understanding that this test requires the uh snapping their fish to generate the traffic right.

E

Not really so this functional part is ready to use PTF. So it's okay, so that's functional there. Where we committed the dash venetics case. It doesn't require it. The like standard, PDF tests test. You can run it on the PDF environment,.

I

Oh that's functional, but but what about the not non-functional like.

E

Yeah for non-functional there was so we are about to introduce Snappy for running line rate Traffic to be able to connect to suffix Network. So that's that's work in progress.

I

I

But are you saying that current all these tests does not generate the traffic.

E

They generate all of them generate traffic. The question only in in which way, because functional tests using the PDF has a traffic generator and they're sending like a few packets, and they do like my personal understanding. So that's the idea is that in the PDF and functional we are going to analyze the packets, so we have refined the packet transformation. Everything is okay, but we are sending just a few packets and in the non-functional the idea is to send the line rate to send millions of packages.

I

Yeah yeah for that one right, I'm, looking at the you know on the scale folder right, so is that a non-functional is that scale test under the scale folder.

E

uh Currently in Maine, if I'm not mistaken, there's there's some examples, so maybe Mirchi can give yeah so we're not doing it. It's.

I

So I'm, looking at this test, underscore PSI underscore we need underscore inbound.py, which is under this uh scale, folder.

D

Go on, could you share a link.

I

I don't know like yeah, let me see if I can you can share I'm looking at this, the this one right so.

I

This one, this one is under scale, folder right so and.

I

So this one, it seems to still using the PTF to generate the traffic right so.

D

Could you share the link because.

I

I uh put in the chat window: yeah, okay,.

J

D

It are you sure, you're in the right meeting.

I

Oh no I should assume so I think I did share in the right meeting. I think yeah yeah, absolutely right.

J

I I don't see it.

I

Oh okay, okay, okay, maybe.

J

I

Didn't just send the window, but.

J

K

I

This test right, so this test, he seems still using the PTF uh generator, not not using the uh the snap interface to generate the traffic.

E

Okay, so that's one is written that can be launched on both like it's supported, something that we showed last meeting so the just uh the matter of the environment. You need to put in the proper Json file. Where are you so.

I

Then, which one of the tests that the user snap interface.

K

I believe these are the one that is being proposed as the part of the pr website Challenger, which.

I

Has yeah this is I'm.

E

Looking at right so which one look at the test uh outbound, but there should be scales because they're, simple and scary, this is simple words. Oh.

I

This scale yeah.

E

Okay and see that it's a little bit different API for running traffic like scale the net outbound flows set config, start traffic and wait for like yeah, so the different API, which is this is fixed pocket next, is fixed. It duration, where, where we send some period of time.

I

So, where is this star? Traffic gets implemented.

E

uh In the uh scroll up in the file list, there's some Dash helpers folder. Oh sorry, Dash.

E

In the in the left pane just scroll up, because there should be uh helper, yeah, Dash, helper yeah, we need to be net helpers.

I

E

E

So that one is for snappy, because I'll start traffic set config it's standard methods, but here we are just here. We configuring flows in the uh you.

I

Couldn't find that the star traffic Master you.

E

You won't find it because it's uh it's based on the from it's from the sub module, because we are here. We we hear defining like in the PTF you you use inside Challenger and the size Rift as a sub module here. So there's no definition of the start.

D

E

Traffic cancels.

D

Yeah, it's part of the traffic generator abstraction inside calendar right.

E

I

Oh so you're you're, seeing like if I find the tier.

E

uh Yeah yeah, you can find it here in the data plane.

G

E

Plane Snappy, for example, yeah. So that's no.

E

And say that a plane- that's one, it's actually in here, inherited from other classes but but like uh you should find it here.

G

I

I

But they snap you right. So how does he know you know he used yixia or not.

E

Actually, that's something that you're defining when you're, creating that uh Json file. So you can, you can set the xnet direct, so so actually right.

K

Now you can see it right, yeah yeah! You can see it right here. If self mode is ixnet, your X.

E

Yeah, so right now, actually the difference only that we have for putting some extension and the rest in the Json file. The code does not is independent from the what the back end you're using for snapping each weekend, tourism.

I

I

So all these codes are under Apache license.

I

I

uh Yeah but but then back to the questions, how is that uh you know test topology being set up? You know, I, understand the you know you, you.

E

Haven't: okay, okay! So like initial question, so there is no rips to set up that topology, so we are dependent that uh it was uh pre-set up. Let's say by the based on by someone else before so the idea yeah. So the idea is that probably Sonic management has an ansible which is doing setting up of the topology here. We have some scripts for doing it, but.

I

Either description of the topology in order to run this test where.

E

Where the topology is defining the way, how to connect to the environment.

G

E

What to expect in the environment? Okay, so there should be Json file. No, no! Let's see that there is idpu client, server, PDF or Snappy two files below below a little bit Yeah. So above about too far, they start inside underscore dpu, client, server.

I

E

Yeah yeah, so that's it's topology description, but it's uh it defines what to expect, but it doesn't use to for uh setting up topology the like expectation, because then framework will know how to connect. So in that case you switch to the sniper. For example, it will be expecting that snap is up and running and it can connect it in the mode in the xac mode.

I

I

I don't know like Alias the CC. There are some kind of uh um not sure what is the area she needs means.

E

Alias is a kind of name just for ref reference, because here we have in the connections you have uh then using you can use Alias. So that's I'm using the, for example, zero one but like it, it could be any anything actually and then in the test. The main idea, actually that in the test code, then you are referring to the port by the teles, so that you don't you you don't care whether this is verse, 1, verse, 3 or like something called uh a and P0 e and P.

E

One doesn't matter for you you're, just using zero one.

I

I

Oh so this this test, you know you, you have both uh PTF setup and the snap yourself. So in the test you will Define, which setup you're going to use and then select the corresponding setup right. So yeah.

I

I

Now, how do I know this test using the PTF setup.

E

uh At this moment, I put just marks, like my mark, ptfc, that by testmark PDF but um uh yeah. So maybe it's like uh the idea like in future to have maybe some smart marks like in the Sonic management that we can just skip if test, if environment is not supported, because, anyway, you have you're providing the Json file. This test button, you are able, you know, at the very early stage, the capabilities of the setup. So if test is packed in, for example, Snappy and you have PTF okay, it will be skipped in simple.

E

Vice versa, as well.

J

So this will become actually that ptf2 topologies that we were discussing earlier. So basically, if ptf2 topology exists, you can run the case if ptf2 doesn't exist.

E

J

Cannot exactly.

E

I

The the ptf2 topology that Prince mentioned right so.

J

But I didn't hurt anybody volunteering to create it. So it's Prince I Prince. You said you don't have time yeah.

I

Oh okay, okay, I.

B

Think we should work with Richard. I can start a meeting, maybe uh gohani I can invite you and um everyone you know interested I can buy. Christina uh Richard basically uh will need. um We need to discuss how to expand and expand the test, but to support the dash portion.

I

Yeah yeah, yeah and also you know this uh traffic generator right so because I think a PDF is only you know doing most like a functional test right.

B

I

B

I

B

A

From the start, yeah.

B

And we are also looking at how to integrate PDF with Snappy. We can get back on that, but overall there are lots of test cases and all the functional test cases we have also merged uh into PSI and they're. More that are getting added for different uh test cases are being added for all the uh areas we need to win it connection, tracking. All that we are adding that is uh yeah.

I

Where are those tests? Sorry I mean not not here right, so you're not talking about here right.

E

So that's you're, looking right now in the pr so like. If for the Intel test cases, you can look to the main, because uh actually they are already merged.

B

I pasted in the chat Gohan.

B

K

B

And we are going to be adding more yeah, no.

I

E

These are Dash; no, no, they.

B

Just got much to say as well: these are.

I

Related to Dash or not.

C

So this is a Gohan. This is the underlay test cases that are related to the dash, also, for example, like Andrea rooting, for example. It's also related to the dash. So it's.

G

C

It's not over, like.

G

With fewer ports yeah, yes,.

C

You you can run the same test cases that you you can run on the any Sony, but still you can run on the AP also, and we selected uh test cases underlying this cases that can be run on the devices. With few, of course, like two four four.

I

Okay, and are we going to add, like a um you know, overlay Dash test is here as well right, so.

B

Yeah currently, we just went over that earlier in the same meeting today that we have the test plan as well as test cases and we'd like to merge, merge that into Dash and then also Port. It uh basically uh merge it with PSI as well.

I

I

I think it sounds like we have many places to host different kinds of tests and uh I. You know it seems like on the dash there's two locations and the side. We have one location and maybe starting management will have another one. Okay,.

E

So that's the question that I actually uh raised a few meetings ago, so we decided to go with the tests folder under the dash in the single location, but like we need to stabilize sexual environment. So like the idea is to have everything in one place.

D

Yeah I think at one point at some point as Dash stabilizes and the apis are stable and they become possibly an official part of PSI. Then the dash framework can just import those tests. It doesn't have to have them duplicated right.

I

So, where, where are we on? That I saw the the header file, at least what we have defined and stabilized right? So where, where are we on? You know getting those header files to merge into the site so that uh you know we, we we, we can write tests, the you know, host those uh yeah, so I'm just trying to understand. You know who who is driving that to get those uh API, sorry, those apis to be to be merged into the assignment Branch yeah.

B

I think Marianne had presented in the PSI community and I think he had raised a PR2. Oh yeah,.

I

And what is the feedback? Is any feedback I mean? If you guys, uh you know, maybe Intel you guys can merge and we can get the which PR.

B

Sure um at the at the review itself went okay, no many hurdles or anything like that. So.

G

B

Of course, a while ago, as well, I can find the pr this.

I

One, oh, it's been cool.

G

I

Maybe this is come here right, maybe you know like uh Intel and whoever you know can get it pulled, and then we can get this merged right. So then we can.

I

We can get the process and later, if you I have a new API as well, which is the you know, discussing the dash community and then option gets approved.

D

Yeah, if we we can treat Ash as kind of an incubator right and then and then promote and so I think. That's the approach we're taking. We went the quickest way possible to make progress and then push things uphill.

B

Yeah we can help with getting this review done, marched.

I

We also have a personal guys and other folks that can look at this pi and the improve and we can get emerged right. So maybe we should set a date, and you know it's p, this header, if I've been dashed Community for a while right, so.

I

You should shouldn't be any. You know. If there's an issue, then we should be raised uh already.

A

So that's ten Gohan. Should we stop here for the day? Oh.

I

Sure sure, okay.

A

Okay, guys, everyone have their action items to go forward into the next week.

D

I think so: okay.

A

D

A

You everyone for the time and the the in-depth discussion I appreciate it. I'm gonna stop the recording.

D

Thanks Christina thanks. Everyone thank.

A