IPFS Package Managers Weekly Sync, 12 Nov 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 📦 Package Managers Weekly Sync - Nov 12, 2019

Description

Meeting description here: https://github.com/ipfs/team-mgmt/issues/1043

For more information on IPFS
- visit the project website: https://ipfs.io
- or follow IPFS on Twitter: https://twitter.com/IPFS

Sign up to get IPFS news, including releases, ecosystem updates, and community announcements in your inbox, each Tuesday: http://eepurl.com/gL2Pi5

A

Okay, now I'm recording this is the package manners weekly sync certain late for November as well. First up I, pianist Dean.

B

Yeah so short version put some pub/sub changes last week. They now need to bubble up through various layers to make them into octopus. There's a couple PRS of the notes. One is for the pubsub value store or the ability to pee pups. A router has some internal custom. Braking API changes shouldn't be any problems for anyone. They had been on discussed Olympia P that IO for a while and no one has raised many issues and.

B

Figuring out how to bubble things into like a fuss and having some minor issues with sureness tests that having some extra eyes would help us and then doing some test ground tests for IPs.

B

That's all for me.

A

Thank you all right, Turk.

C

Yeah so updates on its what this week, I implemented a test ground test which allows us to simulate a multiple, each multiple seat scenario: yeah, that's better and then just doing working through code reviews for deliverance.

A

Okay, I get 1.5 still discussing the spec. Oh.

A

Sorry innovations mount I, Dominic left us on Monday.

C

A

This is crummy I'll pause. He may become there. Do it on the side in the future, but not for a moment.

A

Ip this ad performance.

A

B

They don't move.

A

To a community.

B

So alright there's some consensus on supporting a synchronous data stores, so this means that we'll be able to actually utilize, like you know, Badgers asynchronous right capability. If we want to and then manually decide when or when we want to sync, we can for people who already have badger like ipfs badger going. They can just change it config like an option in the config file, and this should this should do it for them.

B

We are assembling a list of the places that they have stores used, so we can figure out in the switch from synchronous, asynchronous data stores, where we actually need to call listen command worth, noting that we actually have a good deal of evidence that I pivot, even with adding no sync commands anywhere stuff, is not the end of the world, because Windows has been doing this for like since Bajor started, and we have not had tremendous Windows problems so.

A

B

A

Of I mean, if it crashes here in problems, no.

B

A

B

We have like a year of testing, so that's kind of nice by.

A

Accident, so what is we think of time? Islands would be I'm hoping to get this done before lab week or at least have something ready for lab week. I mean.

B

So again the the flipping the flag you can just do now: yeah.

A

But I mean like not like auditing everything and then like making the necessary changes.

B

So I I can I can work on some auditing this week. If you want to prioritize that just a matter, what we want to prioritize I suspect nothing in Lib. P2P cares, if it ever makes it to disk, and so it's literally just IP of us ad is the only place. We're gonna have to flush anything. You have in any IV other side and I'd.

A

Like to gets out of the way as soon as possible, I think it'll like take maybe a half an hour to just grab the codebase, find everything we write it up. The paper parts that she just like: writing it up: okay, but they want to do that. We can actually move forward of us. So yeah.

B

I mean okay prioritizing this somebody I cannot stuff. That's. Thank you. I.

A

Datastore performance I'm sorry go. This is at this point. Basically a performance Oh same thing: I. Once we get the same thing fixed. My plan is to switch the battery. That's the same as last week.

A

Okay blockers asks finalized okay, ours, but that was this is a tripped. Sorry walkers asks.

A

A

B

Okay, yeah I'm when I don't know that I'm, maybe just focusing on performance stuff this week, but I I need a little bit of guidance on the shortest tests, because yeah.

A

You're, just not working I I.

B

Got the shortest tests to run I'm, just like not 100%, sure I understand why they're doing what they're doing, because, like my manual tests, don't seem to be in doings and things and like as well make sure I'm not missing something obvious I.

A

A

Alright, this is the issue where, like you're, seeing an extra PID and list the PRI DS yeah.

B

Could have discovered this doing.

A

That it could be that we introduced a bug in you know in the the it.

B

A

It could also literally be like: did we accidentally upgrade that he achievement your this, because we upgraded the DHT we added much Ottomans dropping code. We could have also entry sometime, in fact that.

B

Is possible, I guess I, don't know I'll to look more into it, but I can look into it later this week or next or whatever in a couple of weeks. If we want to okay, just whenever we're ready to like get this merge, so we can be like IPS stamps go.

A

Okay and I will take a look at that and try to help you with it other than that I think we've managed to wrap up in two minutes, despite starting late late, I'm set for that I will post the link to these notes in the issue. If you want a hat, isn't gonna miss, but it's actually before we move on Derk. Alright, you are heading out later today. Right, yes,.

C

A

Well, have fun.

A

B

Question first for Derek: how is your experience in writing? The test ground plans because I haven't like are using Falak, give us nodes or small things, so.

C

My original plan is useful, like they're passed notes, but then I realized actually Raul pointed out, there's probably better to test bit swap directly just so we don't have.

C

The mornin allows us to control the way that we request blocks so rather than ipfs, which has kind of a fairly conservative prefetch mechanism we can just like, like pushed up through as fast as possible. So it's a better test of its weapon self and number two. Just it's just one less. You know it's sort of more isolated. So if there's any issues, we can isolate them to bits. Well, I.

B

Was unfortunately, I was hoping you were gonna, do all that work for me or doing actually best things like just copy paste and then switch which part of the core API is used. Well, you.

C

Might be able to just use his I think it's felt small, actually, maybe it's Jim, but anyway I think it's called small bench. Okay marks will be APR, okay, cool thanks.

A

All right guys, you see.