IPFS GUI and Browsers Biweekly 2020-05-19

Connecting to the cloud server. That was scary. Marcin Rataj (lidel): Hi, welcome to GUI, and in web browsers. Marcin Rataj (lidel): Bi weekly call for 19th of may 2020 Marcin Rataj (lidel): Same format, maybe better video recording will see that Marcin Rataj (lidel): I see Jessica has the first and probably second items Jessica do want to jump straight to that. Jessica Schilling: Oh. Oh, yeah. All right. I put this in these notes like a week and a half ago, so I didn’t forget and clearly they serve their purpose. Jessica Schilling: So yeah, so I’ve set up a metrics dashboard in County that is basically a comparison of I PFS desktop and web UI on a bunch of different things. Jessica Schilling: Currently does fire off every Wednesday, a very abbreviated version of the latest stats, it does it in an email is like a giant image and so Jessica Schilling: So what you’re seeing is actually only some of the stats, because the giant image with all the stats, which is like completely unreadable.

01:13 - Jessica Schilling: If you are not receiving this Wednesday email. I think everybody on this call should have received one last Wednesday, if you Jessica Schilling: Don’t let me know and I will make sure that you get added to that district list. There’s nothing like totally, totally surprising in there as of yet. Jessica Schilling: The one thing that it does point out is that upgrade pace is very different. If you look at it as a number of installs versus number of sessions.

01:43 - Jessica Schilling: So it’s really about, you know, the people who are frequent users. Jessica Schilling: Have upgraded like most of them have upgraded. I didn’t, I didn’t check it this morning, but the upgrade rate is huge. And then you’ve got all these people who you know downloaded desktop played with a couple of times and never ran it again. Jessica Schilling: So it’s a very different view of those metrics, depending on the parameter it look at it in. But take a look at the county.

There’s a lot of interesting stuff in their 02:10 - Jessica Schilling: Second one is also mine. Thanks. Dietrich for helping proof this for me. I made some tweaks to the install section of the IPS website just based on some of the things that that I saw about Jessica Schilling: Like abandonment rate like retention rate in Jessica Schilling: In desktop and web UI metrics. I think you were definitely getting in this situation, which we knew already have people just downloading this thing. Jessica Schilling: for the fun of it not really knowing what it does. And then, you know, playing with a little bit and realizing it doesn’t serve their need or it doesn’t explain their need.

02:44 - Jessica Schilling: So a small tweak to that I don’t suspect I don’t expect to see a big change and retention metrics, but did change the language on these all section of ips.io just to make it clearer. Jessica Schilling: What does what or who might want to download what. So it’ll be interesting to see if that makes any done in anything, but it was time, those are mine. I don’t know who’s doing the design, meaning Marcin Rataj (lidel): I think it’s me. Marcin Rataj (lidel): Okay, so this is a quick PSA sort of maybe I’ll share my screen. Maybe it will walk this time. Irakli Gozalishvili: Does it please work. Yes. Marcin Rataj (lidel): Yes, yes.

03:32 - Marcin Rataj (lidel): Okay, so I will be very brief, we introduced subdomain gateways in go I PFS opened five and for the next release. We’ve been thinking about switching go I PFS to new key format which is Ed to I think 2225 519 yes Marcin Rataj (lidel): So the problem is Marcin Rataj (lidel): The key is much smaller, it’s so small it can fit in the Civ so we would start in lining keys. The problem is that in mind key is to characters over the limits of a single label. Marcin Rataj (lidel): In DNS. So that’s like a heart limitation of DNS spec and a lot of places actually implement that spec, which means you are not able to being a domain which is which has selected to long Marcin Rataj (lidel): Segment Marcin Rataj (lidel): So there will be a design meeting to figure out what’s the best way to handle that we like, we probably could find a custom representation, just for this specific key. The, the geez here is that we want. Marcin Rataj (lidel): The default IP and s keys.

04:58 - Marcin Rataj (lidel): work pretty well with subdomains, but at the same time, we don’t want to introduce anything Marcin Rataj (lidel): That we don’t want to lick any abstractions from the web browser context down to our stock, especially to the LEAP P2P which should not worry about things like that so Marcin Rataj (lidel): I wrote me more here with some ideas. How we can fix it and not just for that specific key type, but for a longer see IDs. For example, if you use char. Marcin Rataj (lidel): Five 512 ah Marcin Rataj (lidel): It’s super long Marcin Rataj (lidel): And it’s even longer than the key and has the same problem. So we need like a generic fix. And then we may or may not have a specific fix for the keys. Marcin Rataj (lidel): So if you are interested in joining that call out yourself you to the list. And I will send you an invite.

05:56 - Marcin Rataj (lidel): That’s it on this front. And another one is also from me. So a bunch of the web UI schemes got approved by Anna. It’s a part of our work on with a gala on getting Marcin Rataj (lidel): Protocol handlers for browser extensions to the point where they’re much more useful than what we already have in Firefox in Chromium, we don’t have anything. So we want to push boundaries across vendors in a, like a Marcin Rataj (lidel): generic way that can be used by not just I PFS but all the projects. So the cool thing here and I broke it. Did I broke it. Dietrich Ayala: We can see it. Jessica Schilling: All right. See your whole browser. Jessica Schilling: Perfect. Marcin Rataj (lidel): So it’s Marcin Rataj (lidel): It’s so small. But here’s the the web.

06:58 - Marcin Rataj (lidel): And here’s I PFS IPS, and I believe there’s also that somewhere. Marcin Rataj (lidel): It is somewhere. Marcin Rataj (lidel): Like social short we got I PFS IPS D lab. And while we were out this. We also got data, I believe, secure scuttlebutt as well. Marcin Rataj (lidel): And that means those Dietrich Ayala: Is this the first time protocol Labs is listed on that page. Marcin Rataj (lidel): Yes. And it’s listed like multiple times. Yeah. Marcin Rataj (lidel): I think, yeah, multiple times. Marcin Rataj (lidel): So that’s cool.

It means it’s formal it’s formalized and you can use those protocol handlers in places when 07:53 - Marcin Rataj (lidel): It was expected to be registered by Allah and it’s a part of our work because if someone wants to follow up the next step. It’s Marcin Rataj (lidel): white listing those scheme new protocol schemes in the web API called register protocol handler. It’s ready redirect only API right now. But that’s like the first step to get them to work without web blast prefix there. And then the next milestone will be Marcin Rataj (lidel): Taking that this API and making it more powerful when executed in a web browser web browser extension context. So, for example, like the first component would be able to use it to register handler.

08:42 - Marcin Rataj (lidel): In chromium and at some point in the future we wants to have a protocol handler, in which browser extension or maybe a page is able to inject arbitrary Bite, bite responses. Marcin Rataj (lidel): Similar to like how service worker works or we’ll see and but that’s the general direction of this work and registering your eyes is the first step. Marcin Rataj (lidel): I think that’s it for this PSA. Marcin Rataj (lidel): If you have any questions. Marcin Rataj (lidel): Feel free to ask Marcin Rataj (lidel): If not, we can move to almost shipped. vascosantos: Yeah.

09:23 - vascosantos: So basically pierced are related, almost to shift thing. So basically the initial endeavor is completely done in terms of coding Jessica to be vascosantos: Everything is done in merged into just a bit busier than 28 branch is basically includes the address book, it will produce a book and also something that it was initially in the scope, which is the meta, meta data book and including also data persistence for all of them. vascosantos: So basically now. vascosantos: Oh, when a another thing. Basically, we talked earlier about, I think metrics to for these improvements on starting to peers data and then vascosantos: Restart and see our own good more along less would be to be connected to all of the peers, this is not done yet because basically vascosantos: We want to use the desk grounds for this and otherwise it would be just to actually tests in my laptop, but we are still waiting on the support for Jason browser notes for that. So basically I’m kind of postponing that until Google as the support vascosantos: So this will come in. Dietrich Ayala: I would, I would recommend just do a quick hacky test anyway locally.

So you get a ballpark figure because she goes not going to be able to work on that for some time. I think so. You might might be waiting a while. vascosantos: Okay, so I will make sure we we have for the families. So this will come in the lip be zero 28 trillions, as I said, vascosantos: So we must all these and also other stuff regarding Israel is last week as well. Jacob is out this week. So we are planning on doing the release candidate in the beginning of the next week. vascosantos: Meanwhile, we are releasing this in the in the in the liberty side of things. I’m also integrating it in Jessica first. Basically, it’s almost done.

There is a blocker on the bit swap update that 11:21 - vascosantos: Alex is working on and I don’t I don’t foresee any further blockers. So hopefully we’ll get these in the next Jason first release. vascosantos: And then sit for the year start then I also added another points, which is the following. And there is the interval. So basically, vascosantos: The initial plan for my quarter would be there. I would work on this initial endeavor of the pure storage is there now.

11:47 - vascosantos: And then I would work on the specs for the multi other confidence in here, scoring vascosantos: Which basically would be the next step in their career story and which we name in liberty as a pure service to which also go went to have in their side. vascosantos: This would improve a lot. The, the, the dial performance of in connectivity in general of the beers. However, vascosantos: This would take a long time until we get to the finish line and consensus with the Go team and considering that Jacob is with low availability in the browser context. Now, basically I talked with him and I suggested that vascosantos: I should probably focus on something more. We, the more immediate value in the rosters context. And so basically we got a line there.

I should just work on the Hannibal protocol for now and basically 12:39 - vascosantos: I linked denotes on the issue about where is the Olympics pretending for but for those of you who don’t know, basically. vascosantos: The loop to handle, who is a protocol that will allow general pure discovery and we that we can, it can be used for bootstrap purposes also feel Tampere discovery and even the application specific routing. vascosantos: Basically, this would be in a decentralized fashion and instead of us as we currently have the Web RTC star server and web socket start server. The star. The star server and all those things. That’s basically create confusion for the users and vascosantos: Basically are inventing the wheel one sentence again with the minor changes on all of them and vascosantos: Infrastructure for all of them. We basically just went to uniform all of these and use the rely protocol and secrets relate to achieve connectivity and with this I also to vascosantos: Create proper documentation for browser users so that we can finally move on with all this confusion about the star servers and now these words.

Now we set up this and basically have a more friendly environment for browsers. And yeah, that’s it. Marcin Rataj (lidel): I got the super quick question, in case someone wants to read about likely P2P specs. Marcin Rataj (lidel): That to specs, which I just added to our gosh I lost it now. Okay, so I’ve added two links to specs on the B2B specs Reebok one is for rendezvous and one is for relays. Marcin Rataj (lidel): So this will be based on the random one or are those expects out of date.

14:24 - vascosantos: The spec is after that the render for one because basically go is also not implemented yet there is a work in progress, br from visual vascosantos: For a while ago, but I think that, then it shifts to something else. And basically it has been stalling for like a year or something. So I think vascosantos: This will be the first implementation of that spec and if it updates, it will be according to the work that we are doing. And if something needs to be changed. We will discuss with visa and will update. But for now I see that as the source of truth. Marcin Rataj (lidel): Makes sense, especially like the browser context would be the most prominent users of this Marcin Rataj (lidel): Next one is Dietrich Ayala: graduations on landing all that all the piercer and all the all the all the books, the P2P has mentioned books, apparently say congratulations on letting the library. vascosantos: Yeah, it’s done. Marcin Rataj (lidel): I think it’s also like words in Rachel writing in some in case someone is watching.

15:34 - Marcin Rataj (lidel): What is and I think it’s about I think you mentioned that. But there’s a link in the notes about the plan to sunset star servers. So it’s like super long arc from the probably the beginning of the I PFS project. Marcin Rataj (lidel): When we had those like temporary measures for running a professing web browser and we still are using them. So this is the plan to finally do the right thing. Marcin Rataj (lidel): I see refer this back. So that’s pretty good cuz I was running out of stuff I could say to stall. Rafael: Thank you. Crystal. Marcin Rataj (lidel): Perfect. Rafael: Perfect. That’s the current topic. Marcin Rataj (lidel): Shipped it has desktop. Rafael: All right. Yeah. So we should be professor stop with plates dependencies basically at the most important was the the go see your love life but what Rafael: That was the main source of the plate and soon enough. Maybe we’ll also need another updates, because we have an important alto data that a lot of users have been building Rafael: And there may be other similar update tissues. So I need. Maybe I’ll need to dig deeper. Because with little later, but Rafael: We really should fix all the plate.

The next release so users don’t get all over again. So I’m going to do. I think my priority to be to recognition of lead that Rafael: The picture and try to make it working when other presets and also increase the amount of log and ever logging, we have in the plate. Rafael: We can print. How many users likely later on when users are still being prompt plating but haven’t chosen complete yet and all that information that could be worse than you kind of problem. Rafael: So that’s basically it. Dietrich Ayala: Was this year. The first release that you Rafael: Know it’s a second second already did one this year but 11 have fun with little help. Dietrich Ayala: And then we Rafael: Have the employment and may they were doing the work.

And then I deployed this one basically alone, which means since vital did a good job of writing everything not bothering again with the release the 18:17 - Dietrich Ayala: First solo release. Yet Rafael: And and the other. The next topic is awesome. So I can keep it going. Rafael: The file upload feedback in homocysteine and it was good but can only person months I already answered, you Rafael: Your feedback can send you some videos you can see I’m gonna share my screen that’s right now. Rafael: Can you guys see my screen. Rafael: You guys see my screen on Dietrich Ayala: Black screen then it goes Rafael: Maybe I should lower my resolution because I think it’s on. Rafael: arty farty P or it will be a lot more or no. Me just Rafael: Such that they need to be. Rafael: Better now.

19:15 - Rafael: It’s awful and Rafael: So, Rafael: The file upload it cannot Rafael: Actually tell you everything about it because there’s a book. I think it’s the contraction of the bugs. I really need to Rafael: Figure out where the problem is. And thank you for helping with that. I think the problem is involved the Redux Bundler, which doesn’t directly provide feedback when there are changes. So when we have a lot of changes like this progress is being changed a lot of times Rafael: The Redux bundler like Bumbles folder and just provides feedback later. So when we actually want to tell progress for growing there, we cannot do it because it’s having too many changes and the other ones. Also the file.

20:04 - Rafael: Is consuming a lot of memory in walking the credit directly show. So we need change that will make the Roman percent working but nonetheless. Rafael: If I upload the file, it says how many files are important. I think it’s starting to collapse because I already collapsed. If you want to do them. Rafael: And now, it keeps the record here of the files you upload it and find it. I need to think of pain between here and here, because this one as the Smokies of them. This one doesn’t. I’ll show you with megabytes. Irakli Gozalishvili: Yeah. Rafael: Other than that, you can see this, you can collapse and just cleaning await continue to upload stuff. Rafael: We don’t have Rafael: The file upload a picture if I upload a big file like this will just the Rafael: Hanging hanging window right here in facial performance day, we can see that my memory is exponentially growing, we are now I hope I will will Rafael: Get that’s the place to visit. We Jessica Schilling: Had a quick quick straw poll for all you guys because we had a discussion in the thread about whether it was worth the amount of work. Like if you notice manual word. I can’t pronounce that these all have like Jessica Schilling: Well, that should be the generic file icon, the one without any like word lines in it, just a little sheet of paper.

21:54 - Jessica Schilling: The we discussed the amount of effort required and trying to attach the correct file icon I at this stage I’m leaning toward just having that generic blank sheet of paper for everything inside of that dialogue. Jessica Schilling: For the sake of getting this in the hands of people, but since there’s a large number of us in the room. I thought I would see if anybody in this room objected to that. Jessica Schilling: I’m going to take that as a no. Jessica Schilling: So, Dietrich Ayala: It’s a transient dialogue. Jessica Schilling: Yeah, yeah. That’s what I figured.

I mean, like, eventually, like if we were if we were focused at that level of detail, it would be super cool. Jessica Schilling: If it matched the cracked icon, but um yeah I think just the generic folder or the generic file icon. I’ll make a note of that in the in the issue. Thanks. Irakli Gozalishvili: Baby was something Jessica Schilling: Meaning that it is uploaded. Irakli Gozalishvili: And we’re all been uploading Jessica Schilling: Yeah so. So there’s sort of a thing because we had talked about this too. So, in the long run.

We’re going to have a legit like progress bar, but we’re not able, you know, for the reasons that Rafale just went over. Jessica Schilling: But I’m often for for many, many, many, many files the transfer speed is so fast that it would be just like such a blip. So I think, I think it’s a cool idea. But I might Jessica Schilling: I might unless you feel super passionate about it hold off until we do get to the point where we can have like a legit progress bar progress bar. Jessica Schilling: Yeah. Jessica Schilling: So it’s indefinite right now but Jessica Schilling: It’s definitely right now. Eventually it will be real, but it is at least saying things are happening so Jessica Schilling: Cool.

23:49 - Irakli Gozalishvili: By the way, if I may mention one thing is next item. My mind or Marcin Rataj (lidel): I think we Marcin Rataj (lidel): I think we skipped the Jessica Schilling: Companion one this will take two seconds for Jessica Schilling: You all hate compared. Why do you hate companion. Jessica Schilling: And light, all I Jessica Schilling: I didn’t add the things that that you, the changes that you made it the upcoming companion release, but you are you you cut a beta and you’re real close to cutting a stable right Marcin Rataj (lidel): Yeah, I think. Marcin Rataj (lidel): What I did. Marcin Rataj (lidel): Like huge size reduction like Firefox is it was 10 and now it’s two megabytes, which is ridiculous. Marcin Rataj (lidel): Basically I checked which types of phones are supported by Firefox and Chrome and I removed everything that’s that like I just did the best type Marcin Rataj (lidel): A I wish, like I already merged this work on companion preferences that you did and I will cut a better today, and probably a stable later the sweet. Jessica Schilling: Sweet those changes. Sweet. So the preferences.

Now open in their own tab so they don’t look differently, ugly, depending on what browser you use 25:18 - Jessica Schilling: And as of early this morning there now responsive I depending on, depending on what the final cut over date for the docs beta going non beta is Jessica Schilling: I might be able to get improvements to the welcome screen in there. There doesn’t seem to be much point in doing that, if the URLs are all going to change like two days after we do it so Jessica Schilling: I’ll at least at least get a PR in there so that we can we can talk over what goes in the welcome screen, but we may not want to merge that if docs beta hasn’t officially launched by the time you cut your release. Jessica Schilling: Might be a, oh, that’s okay. Jessica Schilling: I rackley it’s all you. Irakli Gozalishvili: Muted Irakli Gozalishvili: Yes, so I did look into the large file upload memory gross issue partially because that is something related to the work I’m also going to be facing with the moving stuff across the workers.

26:28 - Irakli Gozalishvili: Good news is my platform is actually is able to handle it and do the streaming venues is kindly JS API is doing through all the normalization that happens in the pipeline and kind of loses some of the properties. And that’s why it end up surveillance. Irakli Gozalishvili: But it’s solvable. So we can do it. I’m bouncing through this experiment I noticed that I’ve seen might be worse. Irakli Gozalishvili: And processing. Also, I’m not sure how common that would be that as I was uploading huge files a note is that whole size versus whole library like Irakli Gozalishvili: I think in the web era we show two sizes. One is for the total size and one of the specific thing you’re looking at. And I end up with specific thing being smaller than the full size.

27:12 - Irakli Gozalishvili: Which I don’t think it’s about it is the things that makes sense because of duplication, but I felt like maybe there’s something we can emphasize better in the UI and explain Irakli Gozalishvili: Why is it might be a case. Irakli Gozalishvili: Since it’s a positive Irakli Gozalishvili: And I wrote a proposal for know sharing across browser context that I talked about last week, I think, or two weeks ago, I guess it’s in the notes which you can provide feedback you should, because it would be very helpful. I posted that to the discussion group to us feedback from community. Irakli Gozalishvili: It will be in the newsletter and I’ll be talking to three books in textile next week or next week to get a direct feedback from them as I know they’re the biggest or Irakli Gozalishvili: users that use it in the browser. Irakli Gozalishvili: Um, Irakli Gozalishvili: Yeah, so I one thing that I think I also ended up looking into is if we will be sharing I PFS nodes across browsers browsing collects we will have to be sharing coffee. Great.

28:27 - Irakli Gozalishvili: That means two things would have to agree on the configuration. I think this is a problem actually is, that is beyond the browser. Irakli Gozalishvili: Once browser start shipping in natively sin and users of IPF has one be able to say what the configuration, they want to use it will be proud. They’re shipping with a specific configuration. So he’s also the issues that came up Irakli Gozalishvili: Before we textile desktop and radical and web UI certain hardware VO IP address. So all shipping their own I PFS notes. Irakli Gozalishvili: We just conscious to your battery life and make sense to have one configurations that is best for this your environment. Irakli Gozalishvili: And have that.

So hopefully we can get there probably incrementally. I did a preliminary analysis of the configuration options that we have I think some of those could be maybe hosted out into specific API calls others. Irakli Gozalishvili: Maybe not, but we need to figure out something I think there is also wanting to notice that it was not searing Irakli Gozalishvili: Said, Who is in charge is important is and better supposed to be deciding some of this configuration options or is it a user this decision and treating them differently would be important thing Irakli Gozalishvili: So if you care about this and want to help. Please go to the issues that I linked to and that would be a way to help Irakli Gozalishvili: And yeah, I think that’s a probably oh yeah now also be talking to textile three books and other users about this configuration things to make sure that Irakli Gozalishvili: To understand how they use it and what is important. What is not because it is a if nobody really cares about its and it would be easier to change to make it people actually actively using it and it will be a bigger problem. Irakli Gozalishvili: That’s it for me. Marcin Rataj (lidel): Thank you. That’s super useful.

30:23 - Marcin Rataj (lidel): Um, I think you are. But some time ago we disabled window I PFS experiment. Marcin Rataj (lidel): For multiple reasons. One of them being a change of API and like we’ve been injecting the new version or the old version and people been expecting the new the other one looks very short, is that I think we, if we like reintroduce window I PFS Marcin Rataj (lidel): In some form that would probably need to be aligned with this like know the duplication work that you do. Marcin Rataj (lidel): The like node in companion like either embedded Jess I PFS or the HTTP client.

It’s still like the single instance could be like next to I PFS desktop one of options right 31:19 - Marcin Rataj (lidel): So, Marcin Rataj (lidel): I think there’s like an overlap of some configuration options, especially like for JS I PFS which is embedded in brave. Marcin Rataj (lidel): That’s a we’re like an, an example of we have like a custom configuration that’s specific for that environment. Marcin Rataj (lidel): But that’s just like Marcin Rataj (lidel): There’s probably like multiple name spaces within the conflict, like we right now have a single config. And it’s like everything there from the data store network transports, and even like my own addresses which will I will announce myself, we may wants to get some Marcin Rataj (lidel): More graduate control that will be interesting, especially like for embedded J sacrificing brave. So I just wanted to mention that Irakli Gozalishvili: So I just want to make sure that they’re insane, right.

So, is I PFS exposing brave native one is expectations that users will still be able to configure and have different configuration based on those choices or am I misunderstanding. Marcin Rataj (lidel): Yeah, so that’s that’s a very good question. Like right now, it’s an experiment and Marcin Rataj (lidel): You basically have input for the JSON configuration of your note and it’s not 100 100% under control of the user. Marcin Rataj (lidel): Like things like which TCP port is baked it’s automatically like we find a free one so user does not need to like do anything. So it’s like a hybrid mixture. But that’s that. That’s interesting. If we move forward with probably like remove option to like override everything and just give Marcin Rataj (lidel): toggles for a specific things that makes sense. For example, how much disk space.

I want to allocate for this and people may not care about transports 33:21 - Marcin Rataj (lidel): Andrew, do you want to tell us about web UI analytics. Andrew Nesbitt: And yes, just briefly, so I’m, I’m kind of looking into where why people might not be upgrading to the latest versions of Andrew Nesbitt: I PFS bit go I PFS as a library or the demon Jaipur Jess I PFS mostly used as a library and Andrew Nesbitt: Desktop and to a lesser extent web spy but it kind of ends up being a little bit messy, but seems like the analytics is broken down by these like what’s the desktop version. And what’s the web UI version. Andrew Nesbitt: But it doesn’t look like we currently store anything about the go slash JS IPS version. We don’t record that in analytics, which I’m currently making the most horrendous matrix of what versions of web UI we shipped and what versions of go IPS and desktop so that I can then go like Andrew Nesbitt: Because it looks like go I PFS went from 2.4 point four of web UI to 2.7 point two. So you know that anything in between didn’t ship with go IP address.

So it’s a desktop version, which then allows you to work out. Oh, this version desktop ship for this version ago I PFS he Andrew Nesbitt: gets buried. So that would be good to record that so we can work out. Like, what are people Andrew Nesbitt: Actually running inside of these things rather than me have to jump through any hoops and web UI is record is a hash of the like the sea ID. Andrew Nesbitt: In all the source code in no source code anywhere says the version of web UI. Ah, shipping, which is very frustrating.

You can’t search our source repository is you have to look at commits for something that looks like the word web UI. Andrew Nesbitt: A comment next to those see IDs would be super helpful for people to work out what version is going on, or even shipping web UI as a MPM dependency. Andrew Nesbitt: Doesn’t sound crazy to me. But maybe there’s some history that I’m missing. And the other thing I noticed poking around and county is there’s a lot of errors. And I don’t know if anyone has noticed this and gone like it’s actually, it’s not a problem. Andrew Nesbitt: And there’s specifically. Let me share my screen briefly.

35:56 - Andrew Nesbitt: This one, I think a lot of windows open right now and there’s a bit here that the 28th of April was when Andrew Nesbitt: Came out and I guess we also released a version of Andrew Nesbitt: Desktop and it looks like there’s some unique crushes that we’ve seen. I’m not quite sure what unique means, if it’s a person or if it’s Jessica Schilling: It’s a type of it’s Andrew Nesbitt: It’s Jessica Schilling: The actual type of crash. Andrew Nesbitt: So, and you a new name for a crash or some kind of exception that we’ve not seen before. That’s like, I like that we get new ones all the time now. Andrew Nesbitt: Whereas before, although like obviously down here we are getting crashes. They were all the same crash.

Whereas now, they’re all unique crashes, like what’s up with that. Marcin Rataj (lidel): I suspect there’s like a path with maybe user account name. Andrew Nesbitt: In the area ID in Andrew Nesbitt: And then we have these down here. Andrew Nesbitt: So, possibly, people are having a bad time. But we can’t tell. It’s interesting, as well, to look at Andrew Nesbitt: Where was it different page. I was looking at Andrew Nesbitt: Events and the over 30 days I profess in it failed. Andrew Nesbitt: Five know half a million times, whereas like move a file or right to file 1000 times Andrew Nesbitt: This number is gigantic compared to everything else, which makes me think there’s something going on that like is probably not causing most people to have a bad time. But there’s something happening either weird logging stuff that we Marcin Rataj (lidel): That we should check what Marcin Rataj (lidel): Where we actually capture maybe Andrew Nesbitt: Like a Marcin Rataj (lidel): typo. Maybe it’s like actually different action and just the label because it’s like, does not make much sense. I’m Jessica Schilling: Not completely trusting and the accuracy of these event labels.

There’s, there were a couple of other things that were sort of weird as well. Andrew Nesbitt: Okay, next we Marcin Rataj (lidel): Have of labels. Because, yeah. Rafael: If you go to the error page. Rafael: Occurrences is one that’s lunch. I’m starting them all mechanism, the label. We didn’t know it. Rafael: But they don’t need a label about it. Anyway, here on the error page which is weird. Andrew Nesbitt: So, Andrew Nesbitt: Maybe I shouldn’t look too much into that. Andrew Nesbitt: I’m mostly interested in seeing like is there, what’s the kind of like uptake on desktop for shipping.

38:53 - Andrew Nesbitt: Open five and above, which I think I can work out but also making it easy to do that in the future so that we have a good idea of like Andrew Nesbitt: His desktop driving any option or is it just a very small piece compared to other things and then web UI shifting more things people are obviously using web UI when not using desktop. Andrew Nesbitt: But we don’t get much information, aside from the Andrew Nesbitt: Version of web UI and to be able to debug or find out more, what’s going on there. So is that, is it worth me opening an issue to Andrew Nesbitt: Discuss this more on. I’m not even sure where Andrew Nesbitt: I think it’s repo or web. Yeah. Marcin Rataj (lidel): I think it makes sense to open an issue in web UI repo.

Because my understanding is like, actually the most of metrics that we gather from web UI, the desktop just had a 39:49 - Marcin Rataj (lidel): Few additional labels, but still the most of metrics are gathered in a web UI and like when you are itself could like detect the note type, which is it is connected to enter a report that Marcin Rataj (lidel): So that’s will be pretty useful to find the gaps that we like our like blind spots right now. I know that in a DFS desktop. Marcin Rataj (lidel): When you run web UI inside of electron I changed the user agent to the version of the desktop. Not sure if it’s useful but it at least it’s no longer like electronic version. Andrew Nesbitt: Yeah, I can see that in the Andrew Nesbitt: In one of the dashboards, the Andrew Nesbitt: The kind of the other thing that I was Andrew Nesbitt: Looking at Andrew Nesbitt: I’ve lost my last page. I was, I Andrew Nesbitt: Was trying to kind of work out where Andrew Nesbitt: So we have web UI I PFS to I have Andrew Nesbitt: Which is like it’s there. You can go to it.

40:57 - Andrew Nesbitt: And arbitrary run that and it will fail a. We do. We have analytics on that hosted version of the way if you Marcin Rataj (lidel): If someone opted in, you should see that Irakli Gozalishvili: That’s all the UI us, by the way. Andrew Nesbitt: So maybe that’s so. So some of the errors because if you just open that without I PFS running Irakli Gozalishvili: That’s what I was thinking if desktop is not running it fails to connect to the cluster. Nice. Irakli Gozalishvili: I don’t know if Marcin Rataj (lidel): It also Marcin Rataj (lidel): Means we probably need a better label. Andrew Nesbitt: Okay. Jessica Schilling: Um, well. Andrew Nesbitt: That answers all my questions. Andrew Nesbitt: For now.

41:40 - Jessica Schilling: And if you’re gonna if you’re going to open a sort of Jessica Schilling: Amend county collecting stuff ticket i would i would vote that as far as I can tell, we do not have an ad via desktop event in web UI, which I’d like to, I’d like to start tracking and then Jessica Schilling: I don’t see any events for pinning and now it also that’s going to get more and more important as we start thinking about integrating pending API’s would be super cool if we could add some events for for pinning activities both Andrew Nesbitt: And we’re done by a desktop. Do you mean like dragging from someone’s desktop or like add via I PFS desktop. Jessica Schilling: No, no, no, no. I’d be I’d be a your, your iOS desktop. Jessica Schilling: Yeah, okay. Yeah, because we have that it’s interesting to see that, you know, is this like a subset of general like files. Right. Jessica Schilling: And we have that, for we have those events captured in I PFS desktop, but we don’t have ADD via OS desktop, as far as I could find it as a discrete event and web UI. Marcin Rataj (lidel): Something probably can kind of relate it is the fact that right now when we ship with you.

I agree with I PFS go I PFS or Jess I PFS we like hard coded that version that and you mentioned that I’m going to track and that means those sorts of like 43:11 - Marcin Rataj (lidel): Every go I pfs and JS I PFS release ships with hard coded web UI, which Marcin Rataj (lidel): Cannot be effectively upgraded unless you upgrade go up if I suggest I PFS itself. If you run I profess the stop it bundles web UI inside if you’re an IP address companion historically we were embedding Marcin Rataj (lidel): The CD and Marcin Rataj (lidel): We were ahead of everyone else but to reduce like maintenance and different versions, we now use the version that’s hard coded in go I PFS or or JS I PFS Marcin Rataj (lidel): That is me prefacing question like the question or idea that Marcin Rataj (lidel): Historically, I think there’s an open issue to allow Marcin Rataj (lidel): Opening Marcin Rataj (lidel): Web UI from the API port. Marcin Rataj (lidel): Not just that article that CD, but also the DNS link name which is at web UI that I PFS IO, and that got sort of like a power so Marcin Rataj (lidel): Stephen was not comfortable with going ahead with that because it relies on the assumption that the DNS link from the the DNS. Marcin Rataj (lidel): Points at the real web UI. Not like someone could the man in the middle at the DNS level and inject a different version of web UI.

That’s why we don’t have this like rolling release. Because right now, when we Marcin Rataj (lidel): I believe we managed to master dev with you i i PFS IO is updated and if you will. Marcin Rataj (lidel): Merge to release or something like that. The web UI is updated or the other way around. Marcin Rataj (lidel): So we could like we could have a rolling release of web UI. The problem is we don’t want to do that as a default because we want to ship.

45:13 - Marcin Rataj (lidel): Like the version that’s tested with against the specific go I PFS version but I wants to figure out a way to provide an easier way for people to opt into the latest web UI. And I think like in companion, at least we could Marcin Rataj (lidel): Historically we’ve been playing with Marcin Rataj (lidel): Modifying the origin header to pretend that to the we loaded web UI from the API part and that removed the manual step of Setting course headers on go I PFS demon to allow logging web UI. Marcin Rataj (lidel): Allow accessing KPI from gateway port. Marcin Rataj (lidel): I’m just like interested if we because we also are planning to Marcin Rataj (lidel): Like not blinding but we are discussing to stop bundling web UI inside of electron browser and instead I PFS desktop just having like a menu item that for status and files and that link will open in a browser. Jessica Schilling: Which the which the surveys seem to support at this point.

Sorry to interrupt you, but so far that looks like. Jessica Schilling: So he’s like Marcin Rataj (lidel): Actually, I hope to Marcin Rataj (lidel): Have a like shorter description, but it’s actually like a lot of moving pieces right here. And long story short, is that we, we probably will have to figure out a way of making it easier for people to opt into the latest web UI. Marcin Rataj (lidel): release cycle of go I PFS is like six weeks, unless we are like comfortable with aligning with that release cycle. Marcin Rataj (lidel): Maybe that’s a good way, moving forward, but Marcin Rataj (lidel): To me it’s like an open question.

47:07 - Marcin Rataj (lidel): Either we align ourselves with recycles of go I PFS maybe JS I PFS will follow or we may need to find a way be that with companion or not have Marcin Rataj (lidel): This letting users to opt into the latest version. I don’t think it’s possible without companion, actually. So yeah, that’s a hard cookie. Can Irakli Gozalishvili: Ask a question. So I think one problems that you mentioned were a core centers. And that’s something came up before as well. Irakli Gozalishvili: It seems that may be solving the core center might unload many opportunities here specifically like is there a reason why web, you can be justifiable set uses I PFS Irakli Gozalishvili: Internally said we send you could go to the actually could load with a hash of the web UI relates or he could go to the web UI that I command, it will automatically make winters Aveda stash.

But it could like 48:12 - Irakli Gozalishvili: Flexible Marcin Rataj (lidel): It’s actually like it is a web app and you can do that. You can load that hash from the gateway from a public gateway or local gateway. The problem is that if you load it from the gateway port. Marcin Rataj (lidel): It will have a different origin than your API. Right. And then you need to look at those course headers and this is like Marcin Rataj (lidel): Just a problem of user experience. It’s not like we would like, if someone opens, click a link in the web UI inactive has desktop. We don’t want to ask people, we don’t want to greet people with that error, you need to type this stuff in console. Irakli Gozalishvili: To make. So yeah, that was my point can we fix that. I think we should Marcin Rataj (lidel): Be Irakli Gozalishvili: Like if you’re on desktop, we could show the prompt Irakli Gozalishvili: You can click the prompt and like get the course header setup. Irakli Gozalishvili: We don’t use desktop, then maybe something we can figure out for the Irakli Gozalishvili: Team and version is all Marcin Rataj (lidel): Exactly. That’s, that’s what I would love to figure out Marcin Rataj (lidel): How to like remove that manual step, at least for desktop.

I think it’s possible if we pass 49:21 - Marcin Rataj (lidel): The address the potential small problem is that in go I PFS if you update configuration. It’s not like a live reload. Not everything is life love that you need to stop and start node. And that’s like pretty big requires some refactoring in go I PFS Irakli Gozalishvili: Can some holes made me sad because it seems like the course others may not really need to be loading in memory. It’s like you need to know that the request time especially auction. Irakli Gozalishvili: Anyway, if there’s an issue. I think that would be interesting to look into.

That’s also something I want to have regardless 50:02 - Marcin Rataj (lidel): Yeah, I think that there’s an issue, at least for the stuff with cars and DNS link security and upgrading to the latest version. I’ll find it and added to the notes after the call. I think Marcin Rataj (lidel): We got three minutes left, so I’ve, I propose we get those three minutes back. Marcin Rataj (lidel): Unless they’re, like, super important topics you want to bring two minutes before the end of the call. I don’t think so. Right. Marcin Rataj (lidel): Alright, so thank you so much for joining this week. See you probably sooner but on this call in two weeks. vascosantos: If Jessica Schilling: We just want to stay on this call. Go get your coffee or whatever and come back in a sec. Jessica Schilling: I have coffee. I’ve consumed all the coffee. .