IPFS GUI and Browsers Weekly, 16 Oct 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: A Little Bit of Everything - IPFS GUI and Browsers Weekly, 2019-10-16

Description

About IPFS GUI and Browsers Weekly: https://github.com/ipfs/team-mgmt/issues/790
IPFS Mirror: https://ipfs.io/ipfs/bafybeiawg2rlmyfivi23wacsnit2aegleyd2asjmln2lgtd6brrev63g4u

A

Welcome to this week's ghoulia and in web browsers will tickle its 16th of October 2019 and we had an agenda.

A

Let me share screen because the first one on agenda is my demo, so we got the trick. We got hack, two three to add items to agenda, make a quick demo. So last week I mentioned we started working on pre cached web UI, making it available in offline mode, and it's not be ready for magic. It will soon land in beta and three channels, and specifically, we will be able to use this in embedded check with embedded Jessica fest in brave.

A

So if anyone is interested in interested in details how this work he details are here long story short. You.

A

Are now able to.

A

Maybe I'll say that I am now loading build of IP, first companion with this patch, so.

A

The society has companion connected to my local node.

A

However, that's quite DFS: that's not what I want you can see here. It's connected to calculus. What I want instead is to use embedded one so I switched them, but that one it takes a moment to start for the very first time, but you can see it automatically found I found ports for Gateway and API. It discovered my local NGO ipfs and it started importing web UI.

A

What's that what it's really supposed to work? Yeah, it works, that's cool, so what's happening right now is the first time and by the J's API first started. It got the it checked. If web UI is in local repo, it was not so it decided to add it to local cache from an archive which was bundled with IP us companion itself, so that may run for some time.

A

In the background, however, the thing is now when I go to web UI here it should open much faster, mainly because it's already in local repo, so you can see it is probably slow, because this laptop is very slow and I'm running like screen, recording and all the stuff and also like full go idea file. So it's barely working.

A

However, it's super fast. If you run it does not come like a regular machine and all that, like the files and things are now on par with going to your house. So that's good for brave, because we run everything both GPFS and by detected big gateway and IP first companion in the same basically in the same page background web extension page, so removing the need for fetching entire web UI from the swarm makes it much more performant on the first run, but what's really cool is that it makes where do I available in offline environments.

A

So, as long you've got ipfs companion installed, you can play, you can add files to your local nodes using web UI in this interface, even if it's offline, even if it's not connected to upstream internet. So that's a short update. It will probably it will ship soon to above chromium and brave in like following weeks and we will be working to towards like making sure the web. Ui is also available on Firefox, but there are technical details listed on the issue.

A

B

Questions, that's some nice work, so this helps with two things: it helps initial load and it allows web UI to work offline, alik at.

A

Any point: okay, correct and also if you, your local node, immediately- starts providing web UI to other peers on the network right because you get that for free in ipfs.

A

A

Pre cached web UI yep, that's that's it from me. I think the next one on the agenda is IP. First, gay IP, the object of data set I believe it's yours in.

C

Series I, probably I'm, going to share my screen. Even the boo just one second currently I have to give some permissions.

C

Okay, I can I can share it, because I would need to to change permissions. I need to quit this, but no problem. So why do notice that JYP is data? Set, is quite old and I don't mind, updating it I'm just trying to understand everything. That's happening with it: coo coo, coo, books, ooh I, don't know that it probably could use direct recording I like just to know what that means. In this context, I'm trying to understand that I've been searching a bit, but.

A

That's a good question, so basically sharding is a feature of ipfs when you have a very big directory, yeah.

D

A

When you're building Doug, you don't want to build a very huge like level you want to like Charlotte and we I believe Lee. We got something called H a empty charts and that type of sharding is I believe we are basically cutting this huge flat directory each thousand items and making that a separate branch, and that branch is abstracted away on some api's when you are traversing. The dog like high-level LPS, which operate on directories, do not see this additional layer of indirection it just traverses directly and a main example and use of char link.

A

Right now is Wikipedia, because everything in Wikipedia entire thing is under slash. Wiki directory.

B

A

That directories picture that I believe chard. Inc is implemented in Goa, TFS and I, and also like in Jays idea. First, because we are able to load Wikipedia using.

D

A

Fs in brave, so the question is I'm, not sure if the data set is that big, like the data set for for Garrity and.

C

Think it is I, don't know the CSVs and basically the IP in total. It can be, it can reach like 250 megabytes. Perhaps oh, it's.

A

Bigger than then.

C

Did the one we had before, which was only 60 something megabytes? Oh yeah,.

A

Probably probably not worth it, and maybe maybe Coppock cements sort of like a shard sharding on more practical level, so making sure when we like represent this data set on ipfs, representing in a way when I want to make a lego, IP lookup for a specific like IP range or something. I only request this small branch, and none of those like hops on the tree is too big, but I don't think it's a problem. I would just try to replicate like update its using the.

D

A

And to worry about charting, if we like hit some performance issues, I'll.

C

Try to because the new the new T like zeolite is a little bit different under structure. I'll, try to Billy.

C

A

Me me on an issue: if you got like specific questions, because I feel yeah, it's possible that we will need to. Basically you redo the way we represent. This data set an IDF s.

A

We, the problem is like we don't want to like fetch too much when user.

D

A

To to the Pierce screen, and we want, while we do this, we probably should like include ipv6 angry.

D

A

All right so I'm thinkin, that's.

A

Cool distinct, matrix.

B

Yeah I added that now now that uh we're so brave is p0 for the quarter for four browser land and we have p2 for this testing matrix, but the more that we're listening to the core implementations space, the more important this type of coverage I, think, is going to become Co. Ipfs wants to cut a release later this quarter and that's going to be sounds like a challenge.

B

There's a lot of work to do there, so the more that we can get out in front of where they're at and understand. If the changes that they're making are either you know breaking desktop or a companion or web UI, the more we'll be able to help them by notifying them sooner about any issues that we might find. So I think that really bumps up the importance of this and we good we're I'm I'm at the point, give it at the end of the era I'm sure to cut weeks like we have.

B

We have now less than ten weeks before the end of the quarter, which is the end of the year, which means, if the sooner that we can get some of these basic tests up in front of be able to have like latest latest desktop or latest companion or latest web UI running against latest go ipfs even on a even if it was every few days or something like that. So we could know when they're landing things as soon as we can. Let them know things great, the better.

A

I've been sort of looking at while working on on the separate thing. I was looking at how at the test external command in ager. It was audit recently, which is utilized by J's API FS to basically run tests of external project against new patched version of Giuseppe FS and see if those, if the build of external project breaks, when you swap jsapi files swap the release version, they are using with the latest one.

A

So that's something I've been learning created this week for like a Pickers companion but I believe next week we probably could start doing something in this column. I mean it's probably. The first thing we would try is to set it up with JSI givers.

A

The thing is J's ipfs right now is not supported by web UI. So it's like tricky tricky place.

A

The comment itself is like J eccentric, so I need to like understand it and see if we should either extend that and that's add support for like go somehow or create something similar but being like language, agnostic, so I think we will probably get this off the ground next week. I'll, try to allocate some time being Huck our concerns because, like regarding, like J's ipfs support, is it something we I? Don't think it's like a priority. It's like nice to have right now, right, I,.

C

D

A

No like realistically this quarter, if we are about to ship those tests, we'll probably ship those tests before we ship support for JC TFS right. Yes,.

C

I think so, at least on desktop I'm, not thinking about an export, so jessica says right now there are other things to do. Yeah.

A

So it's like usually I, would like ignore jsapi files.

A

However, we really need those tests for brave, so it's not like we I, don't feel we can cut corners here. We probably need to like deliver both this quarter. There's no way around this is.

B

There is there an issue tracking chase, ipfs support, and how do we know how far away we are? Oh, it's.

A

Like just to clarify the JSA PFS support in ipfs desktop it's here. Yes,.

C

Like it's not a problem support, it's really easy to add that support. You just need to bundle gsfs, my previous desktop and it should work, but we have to let either speak between ipfs and jessica. First explain them the differences, why they would choose one over the other and that might be the most difficult part I would.

B

Recommend doing that, but maybe just have anyone not enabled and not allowing that choice yet like do the bundling but not provide any UI or any explainers command.

A

Line parameter just to have like tests in place like.

C

On the configuration file, there's already an option that says the type is go. We just need to change the option and bundle Jessica fest. It was working like that before I just removed it than me, River yeah,.

A

That's probably a good plan just to add support, but don't expose it to user in any way. So because then ipfs desktop would be here and like I, say PFS in brave would be here so like that Firefox remains. But if you have this working, we probably have this as well yep, so I actually probably have either a call or I open a PR with something along those ways. I need to like to see how to accommodate both go in Jess I.

A

Agree like that would be probably my at least mine, my next priority and to figure it out because we really need those tests. Otherwise, we will be like stopping everything and trying to fix regressions and yeah.

B

What I really want to avoid too, is is late, is letting go I PFS know later about problems that the bugs that they might have shipped as well episode, we are, are the fact that that web UI exercises and desktop and companion exercise all these AP is allows us to be an early warning system for problems that they might not have caught in their tests.

B

History is shown that we do discover those issues. Okay,.

A

And like those like early adopters tests, intercept EFS work, I smoke tested them this week, Alex paid me a visit and asked. Why did you break things? So it's very good. Someone is looking at those. The system works system works. We just need to like add more to it.

A

Cool I guess that the status we will try to like kick it off within or like one way like the next week, at least at least, to do any initial research what's needed to support both. Are we able to use edger or we need to create something custom.

A

Go Wikipedia oh yeah,.

B

I had this there's a the Wikipedia anniversary is coming up there. Big one is coming up in January, I, think so kind of like talking about ideas and the fact that we've been talking a little bit about Wikipedia stuff recently. So how kind of sets us up to do some fun things hack projects at lab week?

B

You know Wikipedia Fridays or something that Lytle and I were just talking about our and- and maybe even you know, in terms of like kind of like broader support, ongoing monthly short call of all the related, so people not even I confess, related necessarily, but that are interested in doing this.

B

Work like the opensim folks and the people doing the snapshots, but would love your ideas about fun things that we could do any any things that we could use Wikipedia as a forcing function to fix stuff that we know we wanted to fix for a long time. Anything to support that use case got kind of the conversations we have about co-hosting, also kind of support. This idea of like really low barrier to entry for regular users, to be able to use things like companion to better support Wikipedia by by their own browsing, observing.

A

Yeah I honestly believe that we sort of approaching the point when we I don't need that, like the idea of Wikipedia Fridays or something like recurring call, it could be like initially like internal to people across projects. That's passionate about this, but it could at some point like invite people responsible for making those same snapshots and other things cuz.

A

There is a like there are discussions internal to ideas, how to handle some things, for example, co-hosting spec, which is like NFS based you like regular ipfs node is all you need and there's also like more professional solution when you have like collaborative ipfs cluster, where people that contribute their own cluster nodes, but that, like there's like a higher threshold of when you can contribute. So all those discussions need to happen some somewhere and ideally, we would I think like we will start thinking in Wikipedia.

A

Friday's I'll try to allocate some time each Friday and first step will be to think. How often do we want to meet? Who want? Who do we want to invite like make clean up? The house, like those scripts, need to be like honestly, refreshed, polished, already leches plainly written, but I'm really confident, that's the more we talk about this and people see there's like actual movement.

A

We are able to deliver like actual value because not only make sure like the current version of distributed. Wikipedia is kept up to date, but we can demonstrate things like the strengths of ipfs, just like I said like from DNS link. Ip NS sharding hunting, huge datasets, unpacked Wikipedia is 650 gigabytes in size. So that also is a great playground for experimenting with UX I wanted to contribute, but I don't have six hundred gigabytes of space or how do you like present UX to user I? Had that discussion with Michel last week?

A

How do I, like even present to a user who wants to contribute to Wikipedia? You don't want to lecture, pin Wikipedia because that's impossible. You don't have that much space, and things like that, so it's like entire, like Wikipedia topic, is great, like battleground for across across ipfs ecosystem and I, feel it's a very good story to tell and to and make people excited to work on.

A

Wikipedia Fridays.

B

Do we so you some of this work has already been, is already in the distributed Wikipedia repo, but is there any one place that is there? Is that place where we should be keeping these topics like if we have ideas, if if we have needs honestly yeah, if.

D

B

One overarching strategy issue that we should be using to list all the things that we think we want to do might be related where's, that place.

A

I feel like for discoverability and just working in open. We probably should just create issues in the to be at Wikipedia project I. Like some time ago, I made a first pass and closed and commented on super old issues, but we probably need to do another pass and marriage close remaining issues. However, like that's, that's the repo I feel people are like watching it and we probably should not introduce anything new.

B

A

We could also like things if we decide to make a community meeting. We would just like drop issue there with, like all the links just to like reduce the momentum of the freepo. It was sort of like covered with spider webs, but it's it's getting better. That.

B

D

Cool co-hosting I threw it in there wait. You know we want to talk about it. I have.

C

An idea, yeah question about hosting Wikipedia related things mm. What is done, lazily, co-host Wikipedia and then did they have the idea of letting some pages, because it's occupying too much space right now you only had the idea of removing snapshots, but not pinches I. Don't know it was just an idea. Just a question: I had it's quite simple: we just need to go to the snapshot and do we have to page probably but I, don't.

A

Know like on the technical sides, it's fine like you go inside the snapshot. You could remove any like subdirectory or specific page. It would no longer be like a real snapshot, but the next one yeah.

C

Okay, yeah I mean.

A

Then, when, when the next one is created, that one would be correct again, yeah.

C

A

It's self self healing, especially in a lazy mode. Some kinda related thing is a topic of removal of things from an F s, so I.

D

A

We have an issue discussing QX of the odd somewhere, but a background on the problem is that MF s is like high level abstraction tourism and organized ipfs resources.

A

If you remove something from MF s, it's still cached in your local repo until those blocks are either manually removed or gay collected. So when you remove something from manifest, it's no longer implicitly pinned, but it's still in the cache- and there was like a question: should we either remove when someone removes something from em FS? Should we just always remove, though explicitly remove those blocks, or should we give use a user an option to like remove this and purge from local cache or something like this?

A

And it's like a UX question: would the users even care, or would users know what that means? The the main problem is people add like huge video to amethyst. It takes like one gigabyte of space and then they remove it from em FS and the repo size is still one gigabyte. So that's that's that that's the UX problem.

A

C

A little the rip I profess to remove the blocks if they're not pinned or something like that, yeah garbage collect. We are removing every ring everything that it's not pin and maybe.

A

The question should we like, remove it explicitly or give user an option which is disabled by default right now. The current situation is that nothing is removed, it's in the cache and it will get removed when garbage collection kicks in. But that may happen like next week or next month, depending of the repo size.

C

A

Yep around, like you generally co-hosting I, believe there are some challenges on the US side and on the UX side of lazy co-hosting. So basically, I PFS companion would be an entry point for co-host in Wikipedia in a lazy way and.

A

That's an issue we probably will link it in the notes later, but in the dish P issue we discussed, lazy, co-hosting versus for co-hosting, and the challenge is how to communicate to user the difference which one just specifically in IP, first companion, because that would be the entry point when you click when you are on a distributed Wikipedia, and you want to co-host this, should we just like co-host lazily and that's it or should there be an option to make a full snapshot or it like?

A

If so, how to account for a situation when user has a repo limit, 10 gigabytes and they want to co-host six hundred gigabytes, we've been like discussing things like checking those limits and either enabling full co-hosting only if you have the space, so it's mostly UX challenge I believe because on the technical side, we will basically use existing MF s api's, and we would just run it periodically to check snapshots.

A

Do you see, like any other, like open questions around the classic of hosting.

A

Yeah I need to like grab Eric, I, think and Erica.

D

A

And figured out the you excite of this, but I don't think it's like blocking us from updating ipfs co-host as a library I think that that's more or less unlocked.

C

The P R with respect changes, I haven't changed the whole thing. Sh script yet did.

A

We did we merge the spec updates, no.

C

I was waiting to update the script.

A

Now I think like we should like marriage, spec, okay,.

C

We'll do an yeah.

A

Let's merge, spec and just continue with the implementation, because the UX challenges are sort of like on on companion, site and I will probably be DRI for that, to figure it out, I need to change the UX or around pinning in the campaign companion in general. There is maybe let me quickly show what.

B

What one recommendation that that I'd like to throw out there for for the UX standpoint, is to actually separate these use cases from companion altogether. We talked in Lisbon way back about what it looks like to maybe ship things that users understand who aren't familiar with ipfs.

B

So if you have a router extension that maybe was like companion in the plumbing in the guts but was called Wikipedia helper, when you install it, then you get a lot of these UX questions that can be centered around what the users actually want to do, which is I'm on Wikipedia I want to I want to share this page with other people, and you don't have to learn about ipfs at all, necessarily, but it's using IP FS and maybe there's some branding.

B

That might be fast, though, that no I am serving this I'm doing this on the public network, but it from a UX standpoint a lot of these kind of problems of how do we fit it into a world that is ipfs centric goes away and as a designer you have the ability to say this user wants to share this user wants to save this user wants to read. This user wants to publish again these primary use cases and it really clarifies what we're able to do from a from a design standpoint.

B

Maybe even has has has an admin interface that just opens up web UI Oh for people that wanted dig deeper right, but I would encourage.

B

I would encourage us to when we're thinking about this think about it from the from the point of view of the user, who is I'm, just a person who likes Wikipedia and I want to help yeah or I need to use Wikipedia and I cannot in my country, yeah and not think about it from a like, like let's just pretend all existing UI Justin does not exist when, when trying to come at this from a design standpoint, like I said, have you cases that I read up in the spec bug?

B

All the design should flow from those sets of use cases not from the companion me up today. It's.

A

Highly highly related to the work made by Michelle she's sort of like doing State of the Union of teaming in general, like pinning not in a technical sense of pinning API, we have right now but like making sure the stuff you care about stays around or like contributing your resources to keeping that stuff around, and it's like she's gathering, nodes and other stuff, and this scenario when someone wants to contribute, contribute bandwidth and storage for to Wikipedia is one of those. Maybe like not entry, point I profess, compile its entry point for that use case.

A

Right now, however, like that user story is a separate thing which we we discussed internally and I agree that it should be like discussed, detached from existing interfaces, cause like right now, companion is just a vessel for shipping. Those integrations at some point web browsers would figure out their own way. Companion is just like a template to how this could work. Just let me drop the link to that. Just in the news. Oh yeah, yeah, we'll add it shortly.

A

There's also like a project around language for refresh I sort of started, grouping issues related to that, so that's language, refresh like specific type it has compiled but I believe like redirects. Those are like specific to compile, but things like pinning and the way we provide entry points to web UI. Those are more generic IP first desktop is much better at providing entry points to web UI right now you get a direct link to a specific screen like files or peers in companion got just like one link.

A

But those are you mostly UX discussions, and we have this I'm, not sure if it was mentioned before, but there's there are those tags start to good design which are automatically added to a lot of repose in ipfs and ipfs shipyard. So if you have like a question, you ask question: if you add like design UX tag, it will automatically get added to boards of our design shop, so Eric and others will see it and probably be able to contribute without being like directly mentioned.

A

So that's pretty pretty useful yeah and like the idea around pinning the discussion around being, is ongoing. I.

B

Wonder if there's any there's several high profile websites such as Pinterest, that if then probably UX are around pinning as both language and UX some interaction, design, I, wonder if there's any public research that leveraged I looked in that Firefox UX because they talked about yeah I did there's a couple of different things like pinning stuff on your activity, stream and bookmarks, and things like that had some pin features before, but I couldn't find anything else that was public I know they sort of sites. It's all behind login.

A

I dropped, I, dropped, link to Michelle's notes and I believe there are more notes: I need to find it like generalized user story around on the face, but I'll drop it after this call, because I'm really bad at multitasking all right.

A

Desktop CPU and GPU usage yeah.

C

I'll just add to that someone opened an issue about that and I notice tend to already a new.

A

C

You really won about this, but it's good really.

A

C

Can someone share the screen so because I can't right now, I would need to restore us to zoom I? Can try just take a look at that Activity Monitor screenshot? It.

D

C

Run process is taking too much CPU and GPU CPU, even though I post, it's not close, it's hidden I had the idea we right now we every window the window exists, but it's hidden. So it is really quick when we just when we click on files or peers, because it just opens the other option would be to create the window when the user clicks on on the option on the menu bar, so we're gonna have a rendering process on the background consuming CPU usage.

C

Although, however, that we that we make that time when we the times I'm really confused, is my English for now it it would make. It seems. Clover Bonnie open the web UI afterwards, if it wasn't hidden in the background, I.

A

Sort of wonder cause like in regular Chrome when you have like background tabs. They like some time ago. They start it's like heavily throttling and basically they had a concept of like freezing tabs yeah like a background tab. So basically, if the tab is not the active one, all the like JavaScript, all like timeout set timeouts and intervals and next-next sticks and all this stuff basically stops like yeah.

D

A

Loop stops on until you unfroze the table, I wonder if there's like a similar concept in electron or maybe like an API which lets you like explicitly froze and Aunt Rose I try.

C

To find out, I couldn't I have a new book by the way that only recommended me to my see.

A

C

A

Quite new, it looks like a good book. It.

C

Just arrived today so.

A

Related to our interests, yeah.

C

Let's see if it helps I hope it does, but.

A

I initially thought you meant like ipfs companion, the CPU GPU is high, so I feel totally relieved that it's not another.

C

But I think the main problem is actually I PFS, that is taking most most of the CPU to hear you can distinguish what is actually from electronic bodies and.

A

You, like you, just have a HTTP client. You don't have embedded JSI PFS. Yet, yes,.

C

A

So it's just the client so that that's it easier to reason about.

C

The this person even said that not because when we stopped then the GPU and CPU, that is really well attitude, and it's asking if there's any particular reason at the app consume so much GPU and CPU, and the demand is running since the app window is not open and I. Don't know what to say as yes consumes energy, it's.

A

Not it's not my.

A

It's interesting I will check and try to reproduce this on the next, because it's too may be interesting if it's the same situation across platforms or is it specific to Mac, because there could be something there could be something specific to mark. Who knows right now like just to make sure I understand correctly right. Now we sort of like keep the page in the background running. We just hide it. We don't like recreate and I think each time we open window right.

C

Yeah, that's yeah! That's it, but I think that since we are using redux under something like that, it detects when the pages I don't need this supposed to suspend the activities and remain rear. Enders. Well, it's I know so cause.

A

Like we are talking just to be clear, we are talking about web UI running.

C

On electron, yeah.

A

Whether you ironic on electron and the problem with wet with UI, is that it constantly, even if it's like, if it constantly runs at least request for bandwidth, starts yeah.

C

If you I would like to see, if there's a way to check if well, if I know, if clean window is hidden, if it is requesting anything to the daemon, because if you, for example, is someone opens appears page and then I don't know when it keeps asking frontiers, it really situation and.

A

Sort of like the problem is that, like the way web UI is architecture, is it's a web app that, like queries, our stats API in the background, so on the landing page, you got this graph of like network utilization over time. The problem with like freezing this process is that we no longer gather data to render this graph and we would have sort of like okay, like cache it.

A

The IP ipfs desktop layer and then, when you resume, where do I sort of like populated for the missing history, so that's like a deeper technical problem, I believe but I'm not surprised. There's like there like the GPO thing, it's probably like rendering like this graph I, don't think we are like rendering anything else. It could be useful on.

C

The pure speed rearranging the map and a table- that's.

A

True, but like by default, the page that hidden has this status page right.

C

Only if that was it's the last page, the user visited yeah.

A

So maybe a fix to like solve the GPU thingy is where, when the page goes to the background, to switch to like an empty page which which has no widgets, no nothing. We.

C

Go to above blank yeah.

A

And when you go open web UI again you switch to the previous one just to make sure the canvas does not get you.

C

Know house, if that improves the usage? If it doesn't, then it's not that's not the problem. Yeah.

A

It should fix cuz I, so the similar, like utilization problem brave. There are like other problems around this like bandwidth graph, but if we are able to switch just just like a blank page with nothing because I, don't think that the HTTP request is responsible for this. It's mostly calvess and.

C

Then, like join redrawing.

A

Kleiber II also.

C

I found some issues on the peers. Page I know it's using too much CPU return because we are remain during the math.

C

There's a change in the peers list.

C

Yeah I have to take a look at that. I started it already. Pierce.

A

Page is bit better than it was before yeah.

C

It's better yeah yeah.

A

But like you can eyeball it on the brave it's much better now because for you could see its base, it was basically a choke point. All those like till you get delegated to requests for gay I P things. It was like a huge cascade of requests when you want to talk and Pierce Patchett, better now, I I think like you should try with this empty page and if it does not help, maybe pink me again, we'll brainstorm something. Okay,.

C

I'll, try them out see if it works and monitor.

A

Yeah I have like a similar problem with IP. First companion, we are running a lot of unnecessary updates. I've been sort of slowly addressing kids. We no longer like we withdraw browser action icon. If it's, the state does not change and all those tiny things like a bigger rewrite ahead.

A

Just like people don't really need to see the peer count in real time. I, don't think they just need to see like if Isis online or not. So it's a discussion of TRADOC be providing a lot of like view into the state and like the cost of updating the state constantly, and how much do we want to show out of the box very.

B

First, battery.

D

B

Every every time I did a I did ask in the live extensions channel and from a gecko standpoint, there's not really a good way to profile per extension, the same way that we profile the quarter browser engine, but you can use the extension debugger to be able to do like basic profiling, the same way that you do for for whatever it is. So, yes, some view into into CPU usage there, but not the kind of very low-level a few into what's happening, mostly in.

A

Companion, like most of stuff, runs in in a single like background page, so your base, you see the most of things anyway,.

A

Alright, let me see I believe we are at the end of agenda. Any updates team updates hockey is busier than expected with uni, and you also got a book to read. I'm.

C

Reading two books, this one is new but I'm. Reading the permanent record is almost there. How.

B

Is it worth reading yeah I think so at.

C

Least, for me, do.

A

You have a good sleep that lecture.

C

Lecture is a book or yeah I. Do you know it's not really.

A

Dreams actually.

C

This dream: without that good, somehow there was a storm here. It was really.

B

You know Hugo was telling me that Lisbon does have because we've been having a lot of earthquakes in San Francisco, and he said that that Lisbon does have earthquakes from now and then, but Porto has no earthquakes ever yes,.

C

And we are waiting for a really big earthquake like the one in 1755. Do you know about that? One, no.

B

C

Almost always been god district destroyed, because there was, it was religious holiday and there were a lot of candles. That was not one of the main reasons of the problems that happened after the earthquake. There was an earthquake and a tsunami and then a lot of fires because of the candles and a lot of ways been burned out and like the by Shah, was all built from well yeah. If the parallel streets we have invited it was, it was all built after that earthquake everything that was there was just.

B

That he go mentioned that he said the layout of the city is the.

C

B

Of the earthquake repair yeah.

A

C

A

Candles no candles.

C

Yeah look and the problem was the candles awesome, I've.

B

Been reading a little history in the San Francisco earthquake and it was kind of the same like most people died from the fire, not.

A

Very cheerful stuff.

A

All right guys, I believe we are at the end of this meeting. Let's get back for three minutes of our time. See you same time next week.

C