DFFML Rolling Alice: Engineering Logs: Volume 0: Architecting Alice, 13 May 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Architecting Alice: Volume 0: Context: Part 31: Reset. What is Our Use Case? Implement DIDs to that.

Description

https://github.com/intel/dffml/discussions/1369#discussioncomment-2739704

A

We're going to jump right into.

B

Maybe should I see a lie.

A

Going to jump in, as should I see a li.

A

Should I see a line.

A

I'm just going to make a new file: let's make a new file.

B

Okay, so I realized this tv deck stuff is so the tv deck stuff is likely.

B

Tv deck stuff is likely.

A

Going to be something that happens when we go, it's going to relate to input, origins and.

A

Basically, when your origin changes there's going to be a truss boundary and you might change when you're receiving in within an input network, you might change the origin of an input based off of the provenance of the input which is going to be.

A

Transferred via source currency target currency.

A

C

A

A

Okay, see that last one.

A

A

Conditional offer you want.

A

Target currency and and say that you want, I have currency, you currently hold.

A

C

A

A

Ids yeah didn't.

C

A

Requirements matrix was required to hit requirement services just close cuts. Yes, here the credentials yes or verify credentials.

A

A

Software bid create transaction, send back transaction details. As for the contract check, contract uh sends funds verify funds receive your contract fund, sent to analysis address.

A

So yeah so we're establishing trust.

A

Send funds so that would be.

A

You apply to create transaction.

A

Type contracts.

A

A

Now we need to figure this out for when we go across.

A

Two decks like the appropriate way.

A

And we want to.

A

All right so tv decks like they purple when we want a strategic, find a proposed and accepted.

A

A

Yeah, so basically, if we see this input come in then so if we see an input come in, so we may see an input come in that says, execute this data flow and that input.

A

And the operation implementation.

A

Or the prioritizer.

A

The prioritizer.

A

Would look at the lineage and say: okay, this thing comes with a new system that has to be executed.

A

Input added to network um data flow running version x. It up. um It is dispatched uh wrapped by prioritizer.

A

um And then the purchaser.

A

Checks apparent slash provenance.

A

A

um If coming with a tv, dex um ask reply with bid before prioritizing.

A

Okay, so basically, if you see system context and.

A

So if you see the system context come in and it has.

B

A

A

We'll say: has a method of.

B

A

To be decks, ask.

A

Criteria you currently hold and then you currently hold currency. You want okay,.

B

Okay, so currency you currently hold.

A

So the system context that I currently have the amount of it that I have.

A

Zero, the amount that I want one. I want this to be executed one times conditional offer source currency, the id.

A

Two, the system context that we propose to execute in response which will have appropriate linkage. Data back to the source currency need be target currency, the currency, the customer wanted.

A

Okay, maybe one target amount right or whether not for one okay, so we'll execute one thing right: okay, so this is the offering the id of the context that we're proposing to be executed, and this is the idea of the context that we're we. You know the agent replies and says hey. I can execute this okay, so see.

C

A

And see and ask is the dod of the context.

A

Stand context is going to be executed, uh currency.

A

And the conditional offer is the id of the system context which can be executed by the operation as.

C

A

With additional products.

A

All right, then, the system contacts sure.

A

A

A

Okay, so I think that can let us use that to execute custom contacts so.

A

Okay, so now we can do an ask, we can do an answer, so we should.

A

We can wrap that in a prioritizer.

A

A

A

Accept condition offer.

A

C

A

A

A

And that could be another did dock as well.

A

So this can be implemented in input.

A

And data flows applied as overlays to both part, both alice and bob's,.

A

A

A

A

And input networks.

A

um Okay, so how do we get the other side? Does so hey all right, so one side set an offer. Okay, so it says hey. I will execute this.

A

Okay thing comes back and says: okay, here's a here's an offer, so you can execute this thing so who's going to respond to that offer. Well the input network. Well, we should have.

A

We should declare.

A

The input network should declare what kind of things it might just generate.

A

You know um so, for example, if you're going to define an input, not working, you know your input network is going to be releasing certain events. You know effectively. Now you have to capture those events for your data for your system context to be valid, so we could have it.

A

We could have it so that.

A

The input network must declare.

A

A

That it will just generate at random, and then we have to link those to operations within the parent flow input. Network must have declared up list on instantiation of what inputs it well. Okay, so.

A

So here's a call response, so basically, what if input network now has one more thing so then it has to release event for parent data flow. Okay, so input network must have declared list on what inputs will.

A

It might add at any time. Okay,.

A

Context orchestrator context uh system context: the stem context is not valid unless all these declared buts are mapped to operation inputs.

A

Okay, this is the kind of puts our map to operation all right, then it's valid, and so um one of those one of these inputs might be an input which dynamically.

A

A

Definite occur more to the.

A

Declared list, um okay, so it might add more to the declared list, because what if this thing is running and all of a sudden, it has a new thing. So then it's new things. So this new thing input is basically something that must be handled, so you must have an operation that knows how to handle this input network created new thing because it's been pre-declared and it won't execute if it's not valid, and so it's valid so it has to consume this somewhere.

A

So there's an operation that consumes something that says: um there's an operation that consumes this thing.

A

uh Therefore, the system counters to be valid. The input network to have been used. There must be an operation within the orchestrator context, which will deal with this dynamic. This modification of the input network to maintain validity, checking.

A

A

So now what do we need to do so tomorrow?

A

We need to run.

A

Load, the cli run the overlay data flow, I'm still struggling with how to create the orchestrator so.

A

Past the system context, the system context for the cli is.

A

Okay, so some contacts, so when you define we need to define okay when we run a data flow, we're basically going to assume that when we define a data flow, we assume it executes on the memory orchestrator.

A

And okay, when we define a data flow, they come in from cli. We've got a data flow for the overlay. How do we run it? Well, we load it from a file right, so we load.

A

Yeah, so we like, we load it from a file and then to execute it. We pass it to run and when we pass it to run, it's gonna run in that executor.

A

So we need to pass.

A

So if we pass the day flow and pass it to one executive memory orchestrated by default, if we need to execute using inputs in another orchestrator, we do it from there.

A

Okay, so it just executes the overlay flow loads. The overflow flow executes the overlay flow. Then.

A

The overlay flow executes the regular cli flow, the regular cli flow reads: the arguments executes cv, bin tool.

A

Because this yeah, this is just to see eventual flow that we'll make and then the.

A

Input network so it'll execute the cv, then tool flow and it'll set it'll it'll be a flow that will say if cli and it'll say. Okay, I'm coming from the the top level context, has an input here that was passed to the overlay or the overlay stub and then was passed to this, which is cli, and that way it knows I'm running from the cli. So deployment is cli.

A

The deployment definition, which is linked via input objects instead of definitions now, but to start we'll just do a definition is cli. The cli, args and environment get passed. The overlay data flow, which gets executed within the memory orchestrator, which then executes.

B

The so it executes the overlay data flow, the overlay does.

A

Anything that you know is on start and then accepts as an input. The data flow to wrap, which is the main data flow. The main cli data flow in this case.

A

And then the main seal idea for in this case is this one that we make just for cv, bend tool and that one will say: hey if cli and it'll have to accept the cli, because it's pretty declined in the allow less than the parent one had to use it to call it.

A

Let's say ifcli, then I'm going to instantiate the then run an operation, and in that operation run you know, maybe, with the if it's a cli and it says, there's a coupe control file in the environment variable then you run with the kubernetes.

A

You run an operation, so you do a conditional, and you say if con, if kube control file and if the conditional says you you return two values from this operation which says: if coupe control file in environment variables, you say, if not then execute with memory orchestrator, and then you call your cv, bentool subflow.

A

If yes, then execute with kubernetes based orchestrator. Okay, that's what we're going to do tomorrow.