IPFS Weekly Call, 7 May 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IPFS All Hands 🙌🏽 📞 May 7, 2018

Description

Discussed: Mount on Windows, libp2p-nodetrust, RFCs, and more...

A

To record alright, this is the all ipfs All Hands call for May 7th 2018 and we're stand I'm the moderator. So if you are attending and haven't added yourselves to the heck back, go ahead and do that now and we're now recording and oh do we have any agenda? It looks like we have one on the list right now. It might be a very short meeting, so TJ TV dominic go ahead with your item all right. So.

B

When des our summer is quick approaching and we do not have mount support on Windows, which is going to make installing air-conditioning quite difficult, hilarious, jokes aside, there is no mount support on Windows, not first party and not even third party. There's two projects that I know of that attempts to have ipfs mountable on Windows, but I haven't had success with them and I check on them about once a year, at least so.

B

I just wanted to make a statement of intent, saying that this is going to be something I'm going to be focusing on it's just at least getting feature parity with UNIX on this, so that we can have a read-only so that we can have read-only access to ipfs and IP NS objects through the file system or at least exposed to the operating system as a file system.

B

But ideally, I would like to incorporate some of the stuff that Alan Shaw has has done with em FS, where we expose also a writable route to the operating system. So there's not much really to talk about at the moment. I just wanted to make it known that that's something I'm going to be focusing on. So, if you're interested in any of that, please reach out to me and that's basically, it.

A

Any question is anyone interested in jumping in on helping domine.

A

Okay, that just.

C

Isn't their windows it's a Duncan or Dukan as a wrapper for fuse yeah.

B

So we there's there's two third party of approaches to mounting right now that used, oaken or rather dokkan fork called Oconee, which they they get as far as to mount a drive letter. But as soon as you start dealing with bigger data sets, they they hang indefinitely.

B

The one that was just posted in the chat is an alternative to that when FSP, which is basically our alternative to fuse on Windows, but also wraps fuse. So it's these are the kind of kinds of things I'm going to be focusing on, and hopefully we can start a discussion on this, where it's like what what libraries should we should we be using for this? What is the feature set we want to target for this?

B

Should we change functionality on UNIX and and that kind of stuff, but the the current approach that I'm thinking of is that will likely want to use win FSP for the windows side of things and at least get feature parity with unix, but ideally we'd like to extend out mount on both unix and windows inside of go ipfs, whether or not we should keep that and go ifs, though, or branch that out to a separate repository, is another thing left to discuss as well.

A

Right um I.

D

Just want to say, I actually have some experience. Writing a window windows fuse. This was an open source project about ten years ago.

D

It died, dokin became kind of the de facto like standard, but III well, I won't waste everyone's time on the call, but just really briefly, dokin has a bunch of subtle problems that aren't gonna, be immediately obvious, um just as an example when Microsoft Office applications open and close files, they do really funky stuff like they make a copy of the file and then open that and just just Windows Explorer does weird stuff.

D

So there's a lot of subtleties here, I'm, like 10 years out of date, so I'm not sure how helpful I'm gonna be, but Windows probably hasn't changed. The kernel hasn't changed Washington years so anyway, if I can help I'm happy to have it talk, offline.

A

Cool, thank you Mike, so Smith Dominic's, not alone yeah,.

B

Nice to have some witness support. Finally, just a quick, quick comment on that too, when FSP seems to have been born out of frustration with the fundamental problems of dokkan. So it's pretty immature at right now, but I'm interested in looking at it and seeing if it's a viable option for us going forward.

D

We could potentially make improvements like once you get a Windows kernel, development, environment, setup which is kind of a hassle um but but easier with. You know the VMS and stuff um you can. You know you can do a lot like step through the kernel code figure out where is it um deadlocking and stuff? And but it's it's it's like a serious undertaking, but maybe maybe we should focus on trying to improve when FSP caused oaken's been crappy for like tip 10 years or, however long it's existed, agreed yeah.

B

A

Right, I'm gonna defer this to asynchronous comes, but it's nice to help. People actually talking about Windows support. That's been neglected for a long time working hard to give it some love, so live p2p, no trust! That's magic!.

E

Hi so I have prepared some slides because I have been working on this lipid cannot trust project. My designs to know it and now I want to actually.

E

Wait a minute I have to select a winner to share.

E

So I have been working on this project, IPP, no trust which basically solves the problem of decentralization in the browser because currently manipulate the network is very decentralized budge. The browser isn't so in a normal node. Now the centralized moonstruck nodes, but you know it also uses various methods such as mdns, on gauge key the discard key, our styles or various transplants, CCP WebSockets UDP.

F

E

This means that there are thousands of nodes that can diet, which means that everything is decentralized, but in a browser there's this problem that decides the bootstrap nodes and I was only a cookies circuit which pushes traffic to the bootstrap nodes, and there was WebSockets are, which is the heck that existed before me, loopy, so good. And then there is also wet RTC star, which seems to kill the browser by using too much memory and.

E

Then technically WebSockets could be used, but that obviously does work on any later twin set nowadays, because every website uses HTTP and so I told why not use WebSocket secure, but that would require certificates there. Scientists certificates obviously don't work. So I made this tiny, DNS heck, which basically converts some IP addresses into and DNS address.

E

That later, can we use that the certificate can later be issued for and that basically hex around all these limitations and now notes that run on random machines can acquire a certificate from deliberately neutral server for their IP address and the browser can dial them directly from any website and lo limitations whatsoever are involved, and you notice this college and our flat sucks.

E

So even if the main node goes down, I don't know, the network will still relay all the fruits up messages, so the network can still work, even if the main notice down and thus the browser becomes more decentralized. There is also a demo available on Liberty, no trust on TK / demo and the future plans are now basically dead. I want to make this module more robust and secure, and maybe finish up the tests and I also want them to get deployed and it will appear infrastructure and maybe make it get used in production.

E

So my main problem has been that simply it look, it looks like not so many people care about it and I just want to raise awareness that this thing still exists and I may need some help and getting it to run production because obviously I can't apply myself and I.

E

Don't know if the point- even that's great, so yeah now it's time for some Q&A.

A

Big questions or comments.

F

Of comments in a question but I don't see, fail. Audio events.

F

Cuckoo first of all, thank you so much for the presentation. Machi I think this is a really nice way to like capture like the intent of the project, something that we actually needed for a while as its it is something very ambitious and very new, and this way we all are on the same page.

F

So thank you for that the well so my question is: does this mean that we can dial to any mill behind the net or just notes that are listening on a WebSocket endpoint, but they still need to have a public IP address? They.

E

Need to have a public IP address or a method of making it public using various methods of not circumvention.

F

If you export any of the phrase, letters of Nats equation,.

F

So I'm asking like if, for the cases so a lot of the browser nodes will have so I might try to understand if it is just like a quick setup for people that one to the point GS a professed nodes on a server machine or if this is actually a solution, that for any browser nodes that stands behind or actually for any just vicar. So that change behind the net that so that it can be dialed from a browser. Node.

E

This is actually a solution for all kinds of nodes, but only obviously note that does not run in the browser can acquire certificate. Also, not all matters of not circumvention, work and I think you can P should work, but there are also other methods that use some methods to confuse the firewall into opening connection and I think this was UDP punching or something that it doesn't work, but it works for mostly all other cans as soon as the public. It should work.

F

So interesting one one more question, I said once in your hand, so I'll just continue like in the past. There were problems were, for example, CD ends had one certificate that then they created other certificates at dependent on the first one to give to different websites, and then, when one of those websites were actually hosting some kind of malicious content, some kind of like that bits that had to be blocked in certain region, then all the other websites got blocked.

F

Because of that, and so there was for some time a blackout for many many companies that they didn't realize what was happening is they're like have you considered this scenario where someone just uses a fighter back, it gets very easy to get a certificate under we pedo pedo or some even other domain, likely p2p, not trust and therefore then uses that six tickets to actually do some bad stuff and therefore blocks every single other certificate in certain regions.

E

Yeah I I didn't really think that true, but I think that might be a problem, but it could also somehow be resolved because there are, for example, also other public domains, for example combat UK, which is the common prefix of commercial websites in the UK, and that also doesn't get blocked, and that's also just the domain. So I guess it should be possible to somehow arrange that we simply say that we are not responsible for everything that desert domains do.

D

E

Some I'll pass on the responsibility to the government to block a specific address and not your domain.

F

Okay, so basically more room to explore.

F

Jenny, you have a question yeah.

C

So it was actually it's as stem from the Stop Online, Piracy, Act and so and I think Microsoft actually had this were actually they. Whatever someone was windows licences on some domain and basically the government actually allowed, when Microsoft's to take over the entire domain and actually reroute, and so so any any potentially malicious activity on. If you want to do what is of the p2p or no trust, it's like the government actually will could take down the entire domain.

C

And given the content that potentially would be going over IP, FS or lewd p2p I. Think then, would that's a one point of failure. I, don't see any other way around it, but it's still um could be pose a challenge and also I think the other Matt. My one question is with: let's encrypt: are you proving ownership with just a DNS with a text entry on a DNS domain.

E

Custom web server forbit it results the addresses and also puts in the txt records, if required. Okay,.

B

So I was kind of curious about WD was saying about someone on the network providing bad bits essentially and ruining it for everyone, something that came to my mind was the ability to relay traffic for other nodes.

B

I know this is something that's not or that's being explored right now and I was wondering if you had considered the interactions with stuff like that, like if I make myself a relay and allow someone to use me as a as a point to relay their content, would that expose me to I guess any bad behavior like I would essentially be an open proxy for someone's malicious activity and I'm guessing that that would impact my node negatively.

B

That's something you would considered I.

E

Think that is something that's actually part of peer to peer circuit, because the beauty, no trust itself, does not do any relaying at all. It just provides what's a good, secure connection, but I think it's something that should be considered.

E

G

In terms of the circuits, the at least in that case, the inside the encryption from end to end deduction Krypton through the circuit, so the node that's sitting in between the Box Igbo and actually able to see the data back of all the node at the endpoint actually identified itself to the other side.

G

So it's not it's not quite like torque, oh, but yes, it may be an issue, but it's all like you're, actually seeing the bits and actually happened a bit through like in clear-text you're, basically just like passing like I encrypted before you catch a gotcha.

A

So where can we redirect this discussion to so we can continue? This is a.

A

And then Mike and daveed, you have any suggestions about how to like how this conversation can continue or just leave it with that I think.

F

As always like, it's always good to have an issue like we pick a few projects to keep discussing it just like to keep track of it. We can also just like use that issue to point to the repeated appear. You know, trust repo and then continue there.

A

Thank you great. It makes it so much clearer like what you're working on and why you're working on it and what the implications are. So it's wonderful. Thank you. I just made the presentation in 30 minutes. Well, what is well done is worth the 30 minutes. I'd say next item. Is me it's more of a just a reminder that ipfs summit planning is coming along.

A

Is anyone interested in being involved in the planning for the summit pitch is expressing interest anyone else interested okay, so all right, Kyle, all right, and so, if you're interested check me and wolves, as we start spinning things up, we'll we'll start figuring out how people can help us make this happen and in the meantime, Mike and I are doing a bunch of the work two beads doing a bunch of the work yeah we'll be putting together. Johnny were you? Is it a question or you want to help a.

C

Mode to help I'm actually I'll be in Berlin the week before in the last week of June, and so we are, we said, oh he hard hard on the dates we.

A

Are not fully fully committed, but we are they're pretty firm on shooting for June July 10th through 13th, okay, so it it depends on at this point on whether we get the hotel blocks and the menu but we'll see power source. So it would be in Berlin or her yeah. So.

H

A

This yeah, so the reblued weren't on last week's we're doing two events this year. The ipfs summit, which will be in July in Berlin, is a smaller meeting. That's for people who are actively involved in contributing on ipfs or lib, p2p or PLD. So it's like people who are existing, contribute contributors or very active adopters and then in November and Lisbon will be doing the IQ test conference, which will be a bigger event. That's open invitation or open registration.

A

Anyone can come and it'll be less it's where the one in July in Berlin is gonna, be like a working meeting, um so very heads down get work done, show up with a lot of context already. So so that's like the distinction between the two but they'll both be in Europe one in July, one in in November all right. Anyone else on that I don't see. No, we have multiple screens I'm checking for him. Okay and then I'll follow up with the people who raised their hands all right.

A

Next item yeah, so RFC's I'm gonna be prodding the the working group captains again to review the RFC's that are open. We need to get those merged in order to be able to have the working groups have meetings at the summit because technically we didn't have working groups, don't officially exist, they they're just like a placeholder that we've been using until we actually formally form those working groups and then I wanted to flag this thing of making decisions.

A

So, like there's lots of work happening in parallel right now and I'm, not sure that everyone I'm not sure that everyone is catching all the decisions that are happening so, for example, Rob last week on last week's All Hands announced that he had an issue that had been open for a while proposing how we handled the like dozens of deprecated github repositories in the IP FS board that issues been open for a long time.

A

He gave a final like announcement last week, saying: ok, if nobody chimes in then we're gonna go ahead with the proposal in that issue, and so now in theory, we're just moving ahead, but I'm still not 100%, confident that everyone who has an opinion about this actually paid attention or knew about it.

A

So does anyone have any thoughts on how like what is the right process we can put into place so that we can be confident that people who care about something know that they have an opportunity to chime in and actually turn it in time to be. I.

F

Don't have the right answer, but I have a couple of thoughts and because we also have some like some questions that feel very similar in Jesslyn. Typically, they are more about like kill deadlines and like best practices and or even like API is like how we structure API.

F

As I know, we used our dogs, which typically are things that are not tracked by people's objectives and key results and typically are questions that require like a lot of people as attention to actually paging and understand what is being asked and what is the timeline to take such decision and their implications and so to make sure that, like we actually have an opportunity to check in with each other, when one of those questions appears, we typically tag those issues. Those those asks, those proposals with exploration, like an exploration label and once per quarter.

F

It's when we actually go through all of those issues and like schedule some time to discuss those and then try to make some decisions over them might try to use that focus time to actually review them, make some decisions and then what whatever it's left its left for. The next quarter, discussion, or whatever appears during the quarter, he's put on hold till the next quarter.

F

Of course, like this doesn't work for all kinds of issues, because some of them are very urgent like they need to be, and you know if one two weeks, maybe four weeks time and but for all the other ones that are less urgent. For example, deprecating people's could be a good example, I think it's plausible to say: okay, here's a proposal like no one is listening or getting time to pay. You know what this means.

F

Let's make sure that, like when we plan for next quarter, when we have those retrospectives when we have those sessions, we also had want to review this issue and many others that are out there hanging around. This is like what they've been doing, I'm, not saying that this is like the right solution for every single project, but well see the approach we took for these kind of questions have.

A

You given any thought to which of those issues would actually make sense as an RFC and should actually that we could use the RFC process as a way of flagging things that are requests for comments that will then either be accepted or not. We.

F

Certainly see shoes before RFC you as a thing but yeah like I, guess like what you are saying is correct, like these should get them out through an RFC rather than just like issue the proposal and it's efficient, but still I, think it still applies right. I keep it in with the RFC it'll, be good to try to schedule all of these, like full reviews and decisions to a one point in time, rather than at like being unsure when people are actually paying it.

F

That you're, not like my experience and what I observe, is like everyone's just like super heads down with their key results, making it really hard for people to like shift into a new context.

B

I was just curious here for things like this that really require the attention of either multiple people or the team as a whole entity. It seems like the main point where everyone converges is email, where it's like. We have these meetings, which are optional. You can hang out in the IRC or not. It seems like the thing everyone convenes on this email and we already have team stuff set up where you can just message the list and get that announced to everyone on it and I. Think things like this.

B

That seems like an appropriate use of that where it's like, if you need decisions from multiple people to say like, should this repo get axed or not, you know, I need a decision this week. Maybe that's the place. We should be doing that stuff, I'm, not sure.

A

Keep in mind a comment on my github issue also shows up, isn't you know, and so the people who are only watching their emails can still also and more. My question is more about what are the things we can't ignore?

A

This is an important decision that needs to be made by next time pay attention if you have an opinion, because after X date, the decision has already been made and your lack of chiming in is a favor and like that's not always true with all emails, and it's not always true with all github issue comments right. So how do we? How do we signal natural conversation? That's extra attention to.

A

B

I agree with that I guess. The only thing for me is that, in my personal inbox is that those things are very obviously separated. I know a lot of people here, probably have filters applied for their github issues and things of that, where maybe that gets missed.

B

Alternatively, though, maybe we could exploit that and maybe have some kind of you know special word or something I think it's picked up by a filter and it's we just really make sure that this is something you should see, but honestly I'm not sure about it.

A

So yeah, we should we'll keep that in mind, so the email should go out. Some sort of email should go out about these things and they should have some sort of special term to to flag this, like action required or RFC.

A

Well, alright. So thank you. Thank you for chiming in on that did anyone else have final things who will go to the next one?

A

All right? Oh so this the confusion that, because this bot has been broken for months and is not generating the hack pads for us, if we use a Google document, we can just have one Google document that the that the calendar links to we just put it into the calendar and buy it. It's always the same document, and we just we just put the notes up like this. So just accumulate notes in a longer and longer document, I guess technically we could do that with that hack. That also is just I.

A

Think heck md gets a little bit wonky when documents get really long. But what are people's thoughts on this I also noticed last week that when we were using a Google document, people added more things to the agenda. People were more active about adding to the notes or correcting the notes. So there was something beneficial on the UX side that was getting people more engaged with stuff in that document which I enjoyed. Anyone else have observations or inclinations, Mike yeah.

D

I'm, a big supporter of the model where you just have a a running agenda: Google Doc in the in the invite either I've observed the same thing as you that people seem more willing to just jump in and edit because the date you know they obviously create a really easy editing. Interface I do recognize the drawback that it doesn't produce markdown. It produces data in a completely proprietary Google owned format, but even in spite of that, I I've seen teams work really well under that model.

A

All right anyone else, thoughts on this. There was some plus ones in the chat. For this see what Lars was hearing, he went away. He might have an opinion. Oh they're in here.

H

C

Only thing I'd.

H

Like to make sure is that we still commit notes to the repo so that they don't get locked in some cloud and and to find some way to deal with. The markdown um I took the note last weekend in the Google Doc and then, after what they went through them and converted them to markdown and I, think there was there's a plug-in or extension for somewhere that does it yeah.

D

I've actually I've used that plug-in. um Honestly, it's not perfect, but it will take your Google Doc and export it into markdown and we could potentially even script that so that it's just like every week, we're just scraping the Google Doc. So yeah, it's an option. Yeah.

A

Good, so in the chat, Dimitri is expressing some misgivings doing it should I open and github issue for discussing this, or should we just go ahead with trying the Google Docs and then change if it doesn't work, anyone feel like this needs to be like go through the whole like open an issue discuss it I mean.

H

The thing we do right now, it's broken anyway.

A

Okay, then I'll set up the Google document. Is there somewhere I'll, just we'll add it to the calendar in wall starting next week? We'll use that all right, so that was mine to be, has an announcement.

F

All right yeah just a quick announcement. Essentially we completed the upside to the JavaScript code guidelines and we got included the legal entities protocol. We started like we started the protocol last week and I'm sorry signing winburn players.

F

You can check the link that I put here on the notes to see the full table like including all the modules that are part of the GS ecosystem, total of 70 and right now, I'm really happy to say that, like 41 of those 70 modules have a new assigned wingman trainer, someone that is going to be available, pretty paying attention replying to issues merging pro requests. Doing LPM publishes of those modules for the remaining I will act as the interim weed, maintainer or just leave container but yeah.

F

So you should see a lot of more progress happening on JSON as more people have now the power and responsibility the ownership for more pieces of the project if you're interested in learning more about what does this mean like again, if you check that JS code guidelines micro movement on ipfs community, it's fully described there and that's it. Thank you. Everyone that stepped up to become a wing maintainer of one of the modules of JCPS. It's progressed to share the responsibility and ownership with you and I am really happy like I couldn't be more happy.

A

Thank you for doing all that work to be great mm-hm. It looks like we've hit all of the agenda items. I added, welcome to new people, because there's a couple couple names or faces: I, don't recognize. So if you're new and you've never had a chance to introduce yourself on this call before, if you'd like to, if you'd like raise your hand or say on the chat, whether you'd like to say hello, just giving a chance for people to jump in, but you don't have to. If you don't want.

A

Okay, I, don't see anyone jumping in all right. Well, then, thank you. Everyone for a good call and we're at the were below the 45 minute mark. So so our new 45 minute length they compromise length, we hit it and then there is another call after this one is that right, there's the JavaScript call is after this one is that right to be okay, so some of you will will be hoppy on another call in 15 minutes. So thank you very much and you guys have a have a great week.

A

Okay, come on. Thank.

D