IPFS JS Core Dev Team Weekly Sync, 6 May 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: ⚡️ⒿⓈ Core Dev Team Weekly Sync 🙌🏽 May 6, 2019

Description

https://github.com/ipfs/team-mgmt/issues/650#issuecomment-489540623

A

Perfect, all right, hi everybody welcome to a weekly je s, chord eptein, sync I will be taking the place of Alan today, he's out on holiday, hopefully he'll be back next week with that can I have a volunteer note-taker.

A

A

No all right, Volker I pick you, ha ha ha okay, all right, we'll go ahead and kick it off then. Now that we have a note-taker Volkers. Thank you. Oh all, right and I will start so. Last week I finalized some work to get pull in plex out the door, so that should be ready to go into je s IP FS, so we can replace the PD p.m.

A

Plex and hopefully that probably won't end up living too terribly long with the async iterator work, but that should see some performance boosts because it is a bit faster than M Plex, uh hon dude next message signing in pub/sub, so go live. P2P is going to be enforcing message. Signing they're gonna be working that up the chain. Lytle. Thank you for finding that out and mentioning that to the JavaScript team.

A

So we went ahead and got message running in place and we'll hopefully be doing a release of that tomorrow for json' slush goes back and then also did some CPU testing of the DHT on JSF PFS I'll be doing more of that this week, so that we can finalize what config works best for the HT being enabled this week, I'll probably be working on also doing mess, signing of validation of the signed messages. So right now we're doing kind of a phase release with pub/sub mr.

A

signing so that we can get the signing out faster, but whilst do a subsequent release, what we're actually doing the validation of the signatures, so that we make sure everything is good and then possibly I need to check priorities. But I should we'll probably be looking at natural result and better address announcing for Jay, Haizlip, VTP and I. Think that is it. Does anybody have any questions.

A

No all right, Alan's notes he's out, but he released a new version of the JSI PFS HTTP api. He is working on some things for the interface stream Luxor for Lib p2p, a few other async iterator items for aslo p2p as well. The next this week, he'll be working on workshops for ipfs camp. If you have anything any.

B

Questions for Alan on any of those feel free to reach out to him or ping him in one of those PRS.

A

Dirk not attending still in Australia, but I believe he'll be back next time for in NYC. So we should see him next week. He's finished up the refs local endpoint he's got an RFC out for garbage collection and is looking for feedback. So if you're interested in garbage collection, JSI PFS up over there take a look and provide feedback for him and then he'll be working on garbage implementation. This week, next up Volker.

C

So I I'm still working on the IPL deform at stuff, but slowly it looks good so now, like I, think 80% of the PRS are a proof. So hopefully it's only the EM I believe theorem. One is missing, which is the hard one and main chase. I, probably think, but hopefully it gets a with you tonight, basically in the Australian time zone and then tomorrow, I will hopefully hit M be able to oh well.

C

If an between this, but hopefully this week, I should really be able to merge things, release things and then the whole IPL d-stick is a singing a wait. Yeah! That's all I have.

B

Great questions for Volker.

B

All right next up whoo go.

D

I guess so what I've been doing? I implemented the file Dom API supports for the regular files API on HTTP clients on JFS core, so these the corresponding, inter interface tests.

D

While doing this work, I created a new repo which is still like in review. This ripple basically aggregates some shared logic and dependencies for the BFS ecosystem. We still disgusting if we gonna go with it or just put the the stuff that that's there in the interface repo, but basically they just authority functions that gets they get really implemented all over the place and having a VM there in a single place will help hard.

D

To avoid to others to reemployment because ring commits emoji, because I found out some implementation, not actually the same I found out a few problems with it.

D

Yeah we're still discussing it, but that should be filled out soon and the dumb file API should get merged. Also added note 12 to see I we twist on on the easier other ripples, it's kind of difficult, because one of our dependencies doesn't really run on no.12. So we need to figure out. That's also I did a bigger plate on easier there's.

D

The Avery dependency merged a bunch of a bunch of liquids that were in the bike lock, so major release be out soon this week for Asia, so get ready to do some changes in all your ripples.

D

Yeah I also fix the electron example on GFS. Also added support for running the tests in electron environment. Electron basically has two types of environments: what they call the electron main, it's basically the main process that is a node like process and the renderer process. That's basically the browser, so the public was already open to add support for its I'm, still a little bit stuck with this, because the Beat's, what tests? Actually the big top tests on Jessup EFS repo, don't actually pass because they need to connect to external move and do some with stuff.

D

But everything else is passing. I just need few outs out to change this to to be able to support the electron, and then we will be able to run our tests in electron environments. We will help a lot because we have some issues about it. I also went back to the IP NS resolve, pull requests clean, those a little bit as a few more fixes to align with the go implementation right now, I'm blocked on the under GSI BFS, both requests- you can see there Travis link on the blocks section.

D

Basically, is the tests related to perserve that.

D

Why should not related to my model? If you guys could check it out, it will be extremely helpful.

D

Yeah I know, I'm also blocks on a new go ipfs release, because basically the last ones broke the multiple files app and will rely on dates on several tests, basic, especially on the HTTP client repo, which tests against go ipfs. So we have a lot of tests failing and right now, I have at least two PRS that actually need the latest version of my BFS, but I can't use it because that version breaks all the tests. So it's not easy to handle the test right now.

D

Progressive read out the week, I'll be working on the penis, specifically making the ipfs DNS recursive by default and the ipfs resolve or custom by default, exactly as the latest version of coop ifs. Any questions for me.

D

Go ahead, walk and.

C

I've, just that I know that from the no.12 stuff that M the IPAs Rico doesn't work on no.12, but it should be simply pendency update thing, but I basically looked into it, but then halfway through I didn't finish it, but just if anyone is once ever look at it, it should be for this before yeah. It's one.

D

Of our dependence that doesn't actually support, notes, twelve native stuff, because they change it and either the maintainer for dependency fixed fixes that stuff or we will need to do it or change. Yes, geez.

C

Of activists, it was just the level down or up a dependency which has already new versions of all. Signatures needs to be yeah, love of the changes.

D

Any more questions for me: oh cool, thanks.

B

Mexico next up, Lyle.

E

Yep so apart from holidays, I managed to add even more stuff to the TR, with greater improvements and, in the meantime, Allen merged upgrade to new package name of happy and Joey I. Think so. I realized on top of that, and now the PR is significantly smaller and easier to review.

E

So that would be very good if someone looked at it because it's like hanging there and it's sort of like blocking cutter work that I have killed and built on top of that, I feel some upstream peers that are not really related to Jase ipfs it's, but it makes it easier for Jase I confess to run in basically chromium apps and web extensions in brave, so I think happy one got merged, I did not add links yet, but I'll probably add them post.

E

This call and I got work in progress on getting basically getting a Wikipedia to load using combated gateway in brave, and it's like related to multiple things. Somehow, there's like overlap with what Hugo is doing with adding support photos for DNS link and generally like fixing IPS resolved in Jessica Fez, but also Alex, are that support for Charlotte directories and like I, identified that that works for immutable paths like IPA, first one, but we don't have any story or code that would support IP NS and that's probably.

E

The problem is that if we a lot of people who just want to find call urls to play with embedded know a gateway in brief and Wikipedia is like one of the first like the first choice and does not work right now for multiple reasons like I cannot support DNS link and also like resolving share the directories and picking the correct file to load.

E

So that's like working progress and I'll be continuing that work in this week, I hope to release a new better of companion with about the J's ipfs that is able to load those things. What I'm doing I'm, basically using Jessup EFS fork that has cherry-picked changes from multiple peers until those changes land upstream just to keep iterating and dot. Get this work style and I.

E

Also like when it's sort of starts landing in J's, ipfs I want to add, like interrupts, s4 HTTP gateway, that check checks for all those like other headers and other stuff, because we have like interrupts suit for the API, which is like checking. If we don't divert between J's and go implementations but for the gateway there, I don't think we have like the robust test suit. So probably we need to create one, and this is a like a good opportunity, because I'm already in this and I- that's that's it from me. Any questions.

F

F

Was just checking out the embedded, JSI PF SPR and it looks really cool the DNS link stuff. There's a note about like using the Chrome sockets UDP API is that the plan for like you're gonna issue, DNS requests from the embedded node eventually, is that I'm just curious, because that sounds really cool yeah.

E

So like right now, I think we are using basically what regular JFS ranking web site is doing. So it's using DNA api, v0, dns endpoint at our public gateway just to get it worked, but yeah we have access to UDP like Chrome UDP, sockets, API and that would be used for I believe we should be able to do DNS, lookups and also like local discovery, which is like also using like sub spec of DNS. So yeah, it's cool. This is really interesting. They.

F

B

Cool awesome. Thank you little except Jim.

G

Hi so last week, I remember I did earlier in the week, but I did some more experimentation with the new HTTP signed. Exchanges. I didn't done some work with that last year and I'd gone to Google Tokyo and showed them my experiment of signing content and loading it using a serviceworker and I'd. Let all my certificates expire and so I hadn't really played with it, and so the Google's in the last week had been sort of interested in what we're doing again and it's it's now been shipped into stable, chrome channel.

G

So everybody in the world can use these things now so I I renewed my certificate and I got a a wildcard certificate and they did some experimentation with that and also wrote a little test to see if I could make it work from using a web instead web ad extension instead of a serviceworker and that sort of work. So the sort of neat so I doing it as framed as one of these mini projects I'm trying to organize but hasn't actually been great greenlit. Yet so I haven't put too much more effort into it.

G

But good news is I'm, going to I going to see the spec writer for the HTTP sign, exchanges and web. The packaged in Portland, because I'm in Portland this week for CSV comp and I'm going to meet a bunch of people and so I'll be able to trade notes. All right. I didn't realize it previously, but this is actually Google's core reasoning behind doing all this work is to support the peer-to-peer use case, so I think be very interesting for us to talk to them.

G

So EMP is sort of like what they can deploy with it now, but they're actually very strongly interests in the peer-to-peer issue or how to do peer-to-peer for real like, but it's very difficult issue for them to do, because that they really have to take into account reader privacy and I I dove in a little bit on what the controversy between Mozilla and Google has been with regards to certificates and modifying SSL to cover the peer-to-peer use case and that's unresolved.

G

F

This we called to speak.

G

Yeah they I added some comments on me.

G

The IP IP IP FS in web browsers, is she 121, which is a sort of an epic and to be there's a really interesting, IETF video from last night TF meeting where the Jeffrey from Portland here and when the guys I met in Tokyo are at the conference and they're presenting their stuff and Chris Korell a--from he's the CTO of Firefox and he's also the guy behind, let's encrypt and a lot of the crypto things and he's very strongly posted he he let them have it like.

G

Raising the IETF meeting the certain interesting the website, you can just see, live the debate and I, don't think the current signed exchanges are going to like Google, has purposely put like a seven day limit on the certificates, just because they don't want to make it into the standard. So I think they're going to want to the it's going to be a lot of work. I think the standards level to try to push forward. The peer-to-peer use case learn more about it this week, hopefully.

B

Great any questions for Jim.

A

No all right, we have five minutes left, it doesn't look like there are any cross team updates. Did anybody want to give one.

F

H

Would just yeah I'll just say that since it's written down as being workshop building week for a camp, please let me know if you have questions about protocol tutorials as part of your workshops in the proto school repo, you can find an issue for creating a new tutorial proposal, so you can start chatting about what those sections of your workshop might look like if you're doing that, but don't hesitate to reach out one.

F

B

Okay, poppy probably be reaching out to you this week from the the PDB chat stuff. So why don't you have an update.

E

One update on the signed exchanges is that, in the same issue that Jim posted like thoughts and comments, I also posted on updates that our public gateways now exposing like returning signed exchanges with correct header. So there are no errors anymore in the latest, Chrome and hello. Everyone can as long kill as you like, either purchase the certificate or got scientists range file. You can like play with it and with the chrome, 74 I, think our latest and there's a question in the chat and that's all.

A

And Google will get you after what come officers question.

I

Hi, this is a kuma vez for metabolic I'm sort of an outside contributor, we're building a browser like client for aetherium and with Libby to be, and so my primary question is the async endeavor. It's awesome. It's amazing. It's very exciting. It's also simultaneously absolutely terrifying, because it is, you know, a larger factor of you about 80 modules that is frequently blocking other ongoing development or at least conflicting, with ongoing development and blocked by dependencies.

I

So you know that's kind of a scary and exciting thing, so so one approach I'm interested, is sort of incremental internal refractors into async/await, so like without breaking external API, moving to a single weight- and you know, while this sounds good, because you can merge them quickly, you don't have to worry about like dependency breaking it does.

I

Maybe it moves the base platform compatibility like you're, moving up to no 10 from node whatever or you're you're changing browser compatibility, so I just want to learn more about like how does this approach sound and how are we dealing with this ink endeavour instead of besides, just like hoping everything refractors nicely all? At the same time,.

A

Yeah, so it would be great if Alan was here because I know he has a lot of opinions on this, but we can also follow up in the async issue in JSF PFS, but to target your question on like doing like the massive sweeping changes versus incrementally doing it.

A

So initially we had a conversation, a one-hour video, a while back, and we had decided that for like the smaller repos, we would do the holistic pre-factors, because they're, quick and easy- and not not very painful but I- know you've been poking around in the DHT, and that is a very large, unwieldy thing.

A

That is also moving very quickly now so I'm also on board with making for things like JSI PFS, the p2p switch, the DHT I think those are things that, especially if we're doing active work on doing large sweeping changes is going to be it's going to get increasingly difficult. The further along in progress we get so I think we're going to have to. If we want to do it reasonably we're going to have to do incremental changes to some of those things.

I

Cool that makes a lot of sense me and I really support that. How do we what's the like official stance on like platform, stability and platform support, so.

A

Yeah, so we actually dropped everything below node 10 a few months ago was that last year forget what it was, but it was a while back that we actually officially dropped it, because we knew we were moving to async/await.

I

I

So then, as you may be aware, I do have one specific, tiny PR in DHT and that's just more like a model of what I want to bring to a bunch of different repos and in more parts of the DHT as well. If that makes sense so very happy to contribute small incremental factors towards async perfecter.

A

Yeah, if you want to reach out in the JSI PFS issue, we can follow up there and alan will be back tomorrow, so we can sync up again there too. We appreciate all the help with that and all the help that you've already given to the p2p and ipfs.

I

You got it great things, I'll pick it up. There.

A

All right, we are it's time so I'm gonna, let everybody go if you need to go, feel free to leave hey.

D

Wait bye, yeah.

A

D

I can can't go to the gold meeting. Can you sink there about making a new release of? Why PFS? Because of the multiple files thing that's are, is broken in the current one, because it's blocking us badly. No, just that! Okay.

J

I will I will call it that that tiny thing, if someone else also wants to hop over to go, be first call it's starting right now and can continue to make that point even more eloquently: Thanks, okay, bye, everybody.