Baseline Protocol The Baseline Show - Office Hours, 26 Aug 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Video 1 of 6: Baseline Protocol v0.1 Messaging

Description

The Baseline Protocol team takes you through the code of the v0.1 update.
Next video: https://youtu.be/lsZQwiE2glA
Go to https://baseline-protocol.org for more information.

A

So welcome this is the baseline protocol version 0.1 review session with kyle thomas is just myself and kyle. Today, john walpert, and we're going to be going through on uh walking through the code walking through the versions uh 1 0.1 um stack and with that kyle. um uh Let's, let's get to it,.

B

Messaging and types are sort of a bit intertwined at the moment.

A

B

um So messaging follows the pack the the pattern of having a provider's directory. um It has an index that has an imessaging service in it. So basically, uh this is a very generic interface that defines a set of methods that you would want will want to implement to support any any any other messaging provider we have implemented. uh If.

A

I was um say: yeah say: let's pick uh anybody other than that, because nats is going to be in the uh whisper there. What's another good one. What's what's one of your favorite.

B

Rap rabbit cube.

A

Yeah, of course, so if I were rabbit mq- and I wanted to add my service to this, what would I do yep.

B

You would go up into the messaging package in the in the repo, you would go into source, you would go into providers and you would basically, you would make a new rabbitmq.ts and then you would write the you would write the code and it would show up.

B

You would want to create a new directory, I'm sorry yeah and then call your new new directory called rapid mq, and then you would have an index and like a rabbit and a.

B

Config we've got some whisper, uh we may.

A

B

A

Not advising using whisper right, we did whisper.

B

Yeah we considered whisper to be deprecated already. We may well just drop it out of the release altogether. um Yeah, no.

A

It was, it was easy to use because it was. It was in gas at the forefront.

B

Yeah, it also made sense because the radish 34 uh sam did a good job of implementing the the interface here kudos to him for that, and then uh it was basically defecated now, but it was still okay.

A

Yeah there were some pretty interesting ideas in whisper. I mean you know it may not be. You know for the future, but uh yeah.

B

A

Pretty interesting.

B

Ideas in there sure well, yeah, no, that's fair! um So basically you know you have, uh you know you're, basically, just loading a factory and then that service from from uh from so let's talk about the mass provider, you know in this context.

A

For people that may not know why we quickly make sure that you know what's the 30 seconds on that.

B

Yes and that's a an open source um cloud native project, uh it was started by uh derrick carlson um back when he was working on cloud foundry. I believe.

A

He's in the he's in the baseline community right, I've seen him.

B

Yeah yeah absolutely yeah yeah, so so, um basically that you know that's got a good pedigree uh coming from you know in the cloud native um uh cloud: it's a cloud native project or the cloud native foundation. uh It's got some, you know it's got some uh some, some really it's really performance. It's been it's it. It's a it's sort of a flexible choice. It supports kafka, it has kafka, um uh binding extension.

A

Wasn't it yeah? It was owen connelly, a guy that works for provide now uh who who first identified nasa's a good candidate for this right just to just to give credit where credit's due.

B

I will give oh and credit for um for the echo protocol the idea behind that. That's a great idea, um we'll take more on that later. Oh.

A

B

Kafka bridge, just something to be aware of here: nats- does have extensibility into other. You know other ecosystems.

A

What do you mean by that um kyle? If you don't mind me uh interrupting there.

B

I haven't used the bridge, yet this bridge yet, but what it means is essentially, uh you can run that's on one side and you can run cache on the other and it should work.

A

Okay, but it was I I got that, but but specifically say you had three. You know eight companies a b and c that were baselining right. They were in the org registry and they had set up some kind of set of processes. uh Workflow and they've got everything set up. uh Are you saying that two of them can be on kafka and one of them could be on nats or.

B

But the reason why that's again not to shield maps over and over again, but uh its architecture is more peer-to-peer than other it's brokerless, and so it lends itself to creating a mesh network uh which is exactly what we need.

A

So I think that's pretty understandable uh yeah now, where do we go from here.

B

So that's uh here's. The here's. The nas service implements that imessaging service interface that we we discussed. um You know connectedness, it's just very it's very generic. It's not specific! It's you know it's a it's just a way to um pipe messages that you receive uh on your on your on. Your stack uh into you know, dispatch it into like an asynchronous message handler more or less so you could basically pull in a protocol message uh and then, for example, uh you ingest the protocol message and dispatch it into the baseline process.

A

So so for this one, why don't we talk a little bit? Maybe five minutes worth of just talking uh quickly about?

A

We don't want to go dive deep on on that's I mean we could do another session with eric on that and that'd be cool sure, but uh let's talk about a couple of edge cases or edge failures right because with messaging, it's all about that, so we'll say: you're baselining, a purchase order uh off of an msa or a master service agreement, which is our canonical use case or alice and bob.

A

If you want to talk about that, I know that's your next demo that you're going to talk about um so you know, you've got this workflow step and you've got to get it confirmed on chain uh before the next workflow step happens. Right, yep, uh your your counterparty's servers down uh right, yeah, they're, endpoints, not only fun, tolerant but but long periods you know like I know I remember uh uh one of the things that the court of flow guys are really proud of. Is that you can like keep a thing.

A

You can keep a listener hoping for like a year and it will elegantly yeah. Yeah, you know, because there's cereal.

A

Yeah, that kind of stuff.

B

Yeah so the way, the way that uh the way that there's a good pattern for that um you know.

B

Messaging services in general, or good, good, back, plane, control, plan type of type of you know those are good use cases, and so because you're, you know, you really want to create a pattern in when you're implementing the protocol to um dispatch messages in an asynchronous way uh and do them in an atomic.

B

You know item potently such that uh any message that you know doesn't get um re-delivered or any message that doesn't get acknowledged is essentially there eventually will time out and be re-attempted and then every single message type that you're implementing can have a different.

B

You know a different time at a different ttl in that regard, when you send a message like if you dispatch a message through the messaging service uh using the messaging pro, you know the messaging package, what you would really expect is for when it's delivered- and it's you know when this, when this side of the wire gets it, you would expect it to be acknowledged over.

B