Jupyter Community, 27 Sep 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Jupyter Community Call - September 27, 2022

Description

Recording from the Jupyter Community Call in September 2022.

The notes from this call can be found here: https://github.com/jupyter/jupyter/issues/663

Read more about these calls on Discourse:https://discourse.jupyter.org/t/jupyter-community-calls/668

A

Awesome, thank you so much for that again. So welcome everyone. This is the September 2022, Jupiter Community call I'm so excited to have you all here. My name is Isabella and I will be your host today before we start there's a few things. I always have to say so one. This is a Jupiter Community event like any other Jupiter space. We are held to the code of conduct. That does include me if you need to review that you can find it at any time at jupiter.org conduct.

A

So that's that two, you know this call is being recorded, but whatever that means to you just be aware and three correct me if I say your name wrong, I'm trying to be better about this, but I can't get better and respect you if I don't know so great. Those are my three things. If you are new to this call, what this is is kind of a big cross, sub Community right, there's a lot of sub communities within Jupiter.

A

It's our meeting where we all show show up share kind of like a show and tell like this is my favorite thing. This is this, so it's one of my favorite calls I'm really happy you're all here today and what we are going to be doing is following the agenda that I linked earlier. Please sign in if you feel so inclined, and that means we're going to start with the short report.

A

Switcher items without discussion needs and then we'll move into the main agenda items I'll be taking some notes and keeping like if we link things in the chat but other than that. You're also welcome to take notes, as you wish, so without further Ado. Hooray, let's get started. I am always so happy to have things on the agenda. So we'll start with the short reports celebrations and shout outs, we have a shout out for Jupiter light. I, don't know who's taking that foreign.

B

It doesn't look like any of The Usual Suspects are on here. Are there.

A

Yeah I was gonna, actually check the calendar event and see if I messed something up, but I always try and join from the calendar in that in case, I have a weird link somewhere. So let me check.

C

I didn't write that but Jupiter light is awesome.

B

Yeah and the I know Nick and at least when I've talked to Nick he's been working on widget support for a while and was really excited about. It, so seems, like seems like a big big deal, especially after the uh 8.0 version or 0.8 version coming out. You know it seems, like widgets are getting some new life now.

C

Yeah, so I probably would just eight just came out. That's right um for people who aren't familiar with the Jupiter light. It is a version of Jupiter lab and a version of Jupiter notebook that run entirely in your browser.

C

So the kernel that you use runs in the browser and there's a few kernels that we support there's a JavaScript one, which is a natural fit for the browser, but there's also a web assembly um python, one that comes with a bunch of the libraries you're already used to and so for purposes that are not compute intensive like if you're teaching a class or if you are just using it uh as a scratch pad or something it can be hosted on plain old static server.

C

It can be sitting on your machine and you just double click. An HTML file, even but I, don't think we're shipping it that way right now, and it requires no particular back end and it is a Miracle of modern, uh different threads coming together right, because these are languages that were not intended to all run in the browser, but because of llvm and because of webassembly. You can now Target the compilation of many many programming languages to be webassembly and to run in the browser and it's cool.

C

So if you go to try.jupiter.org and you try Jupiter lab you're actually trying the Jupiter light version that isn't hitting any back-end services or running python in the back end.

A

Yeah, that was a great explanation. I.

B

I I've gotten, you know, you know slowly I've gotten over the magic. The one thing I can't get over is how RTC works with Jupiter light. That's just the the collaborative flag with Jupiter light is just like game. Changing future.

A

Rtc being real-time collaboration, if anyone needs that.

C

um Yeah, so that is itself pretty magical. There's a concept called the web RTC where you can treat different people hitting a web page via the browser as peers on an ad hoc network where they communicate with each other and they send patches to files that they've modified. So say you type the letter A inside a notebook.

C

It will generate a patch saying in this notebook in cell five somebody typed an a at this location and everyone else will receive those patches and reconcile them to show the same notebook, and it can all happen again without the intermediation of a back-end server. It's pretty wild.

A

Great and so can I just check with this release, then support for iPod, widgets 8 is a big deal right, because that wasn't supported because we've been able to pull a lot of things into Jupiter light. But this is a pretty popular one to be bringing in so um I'm glad this was left here, even if we don't know who left it, because that is a great announcement to have.

D

uh A very quick question about the Jupiter Jupiter light: would it be possible for me to like connect, connect the kernel to a um let's say, you're kind of running on on my remote virtual machine or a Computing cluster here, because because the way I see it is like it could be reducing the time to load it. You know in front end and also could be like improving the user experience a lot.

C

So uh I, don't think so, but I think there's versions of remote kernel functionality and lab itself that might allow you to do this like a remote kernel kernel.

C

Basically that allows you to do this, um but as far as reducing the load time, it's actually kind of um it's heavy weight in the browser, because, if you think about what it's actually doing, is it's hosting multiple virtual machines, one to do the python work, one that was already there doing the JavaScript and whatever other languages are compiled, so loading Jupiter light despite the name, is not actually lightweight it's stupid or light because it doesn't have a background server.

C

But actually you end up downloading a few megabytes of code and byte code to run inside your browser, so the load time, for it is not faster. It's just that the requirements of your machine are lower.

D

Gotcha gotcha, thank you.

A

Yeah any other questions or discussion for Jupiter light: okay, okay, um we can always return to that. Does anyone was anyone so inspired for their own short report? Shout out I'll be quiet for a few seconds to let you decide.

A

A

Great, no, that's uh I love getting this news. It's a good thing. I almost feel like we should have the the short reports like somewhere being like hey. These are something someone decided to tell the whole Community. It feels like there's a little shout out there of itself, but yeah with that we're going to move on to the main agenda items as we talked a little before this call right now we have a main one.

A

So we have plenty of time if someone's inspired to do something after uh ask questions, share things, that's totally open, so we will see, but as we are getting started, I think this is ipy vizu story.

E

Yeah, that would be it yay.

A

E

Take it away, hey Isabella, uh thank you and uh hey everyone, uh I'm glad you're. Having me, this is the first time I've joined the community call and actually I just realized. There is such a community around Jupiter, it's so easy to join. uh So thank you very much for having me and uh yeah I plan to give you a quick intro of of our new tool that we recently built and released for jupyter, Notebook and actually other notebooks, that utilize Jupiter's capabilities and, and it also runs in Jupiter light as well.

E

So uh just uh it shouldn't take more than five ten minutes and I'm going to share my screen. Since what I'm going to show you is visual I will use zoom's opportunity to optimize for video, which means that you're going to be completely gone from my screen and I'm going to speak to it. So if you don't hear me or anything like that, just just tell me: I can at least hear you.

E

um So basically we are an early stage, startup in in Budapest, Hungary called vizu and uh well for whatever reasons that we don't have time.

E

For now, we built a generic data, visualization engine that generates all the different types of charts and transforms so like animates between these chores that are described on its interface uh pretty nicely as you can see, and we were looking for different use cases for this technology, so obviously the the storytelling use case is really uh strong uh like, for example, these are just examples of a four charts showing the same data, but you don't know the connection right away, whereas when we just connect them via animation, everything becomes uh really like obvious for the audience, and um there are.

E

There are actually multiple uh use cases we already identified. We feel that really it's like uh the animation. It's like turning uh these static images to sort of movies. You can show the components of composite index or composite chart. You can show changes over time like with the uh racing bar charts, there's been very popular recently and basically, you can turn a single chart into a story itself, just by showing different uh components and and showing uh the same data from different angles uh to to let the audience in.

E

So um we identify the problem uh that as a startup, we wanted to focus on, which is those of the data scientists that need to deliver uh insights to business stakeholders I'm not going to read out what's on the slide, but basically it seemed to be uh like really time consuming from the data scientists perspective and importance from the business perspective, whereas still things like business owners still struggle to make sense of these.

E

Actually we just the research on uh via LinkedIn poll and more than 700 people voted and said that, like more than two-thirds of data, scientists have to present the something to business, stakeholders at least once a month. So it's it's a general thing. We built our technology in an open source way or engine is in C plus, and we with webassembly.

E

We built the JavaScript library, that's released for web developers in September a year ago, and then this March, we released the notebook integration for data scientists and we also added some new extent that I'm going to show you that's iPad is a story basically, and we managed to uh like. Because of this whole embeddable approach.

E

Umizu and iPad is a story. Are available in all these platforms that you can see the logos of on the screen so but without further Ado. Let me show you how this thing works, because I think it makes much more sense uh it you can install it by a PIP IT Works nicely together with pandas we're just going to load. I mean it's already installed here, so I'm not going to waste time with that.

E

This is my machine and it's a vanilla, Jupiter notebook I just hacked the CSS a little bit so that the output appears on the right side so that you can see the code and the and the output on the next like side by side. The data frame we're using it's from an imaginary company that has some regions some products, some sales and revenue data and first we're going to so build the first chart with that with ipivizu we set the height and the width and we add the the filter.

E

So we added the whole data frame and we just tell if I visit to only show the records where the product is shoes, and we just set the chart up by adding this data series to the X and the y-axis and tell it what should be on the label and what it should be colored by and be at the title of the chart here and, as you can see, we call the chart animate method here and when we call that again and basically I'm just changing the filter here to include handbags as well, uh then ipivizu will just animate between these states.

E

So you know it's really like straightforward, which has happened, that we added the handbags I'm going to do the same in the next phase, adding everything but accessories, meaning gloves, appear and then at the final stage of just constructing this chart.

E

We all at the we had the the the every product here and now so like for one thing that we can do here is to switch from the sales uh to the revenue, which is just by switching the revenue series on the y-axis also on the label, and uh by that uh the animation shows this transition, uh and so you can see that.

E

Obviously there are big differences in the number of items sold and the revenue that they generated, and you can also, but just by calling the animate method twice in this case- change the orientation of the chart and aggregate uh the uh the revenue by product and again for the viewer. It's it's pretty obvious. So this was the version that we released this March and actually the the feedback we got from people using jupyter notebook was that this generic logic of constructing the chart can be a little hard to comprehend.

E

So just adding series to accesses and channels such as the color, Channel or the label channel uh people are more used to starting with a type like a bar chart. A stat calling chart a scatter plot whatever, and the other thing is that this is. This only allows one directional use.

E

So if I call this cell again, we will end up with something really funky and you might not know what's going on here, and so it's it's not uh so easy to use at this point and if I can move the yeah so there's we we try to address these uh concerns and- and you know, work towards something that that people um like better so first, we introduce presets uh and here I'm, going to use a different CS3 that contains data about the sales of different music formats uh in the US, from 1973 to 2020 and, as you can see, we just set the year so that it's a string otherwise I previously might comprehend it as a number and that's not something we want and uh yeah.

E

We said the style here. There are plenty of ways you can set the style and actually be even animate between the style settings. If you want that to happen, and now we we call so we construct the chart and we call the animate method once and now we have this stream graph. That shows uh the sales of the different music formats like CD, with this yellow thingies is the biggest and, as you can see, vinyl was was pretty huge back then, and things like that, but here I still use this generic logic.

E

That has like six lines of code, and now you can just uh use the config.stream so tell if I visit that you're constructing the stream graph and then the the syntax will become much more cleaner. So we will know that year is on the x-axis revenues on the y-axis and we need to stack the data by The Format, so the music format. So if I call this, we basically end up at the same chart that we had before, and the other thing is that we cannot really read the numbers or the years on the x-axis.

E

So we add a little event that will only allow the numbers to appear if they are divisible by five and as you can see it already updated. You can actually hack the hell out of iPad visual with these events. You can do whatever you want, basically with it, and so just a couple of things that you can do. That shows the power of animated charts is. If you switch on the Align stretch, then then it will show you.

E

The percentage of shares of these formats in this example we're just going to switch back to the center alignment for the stream graph.

E

You can split uh the the data being shown, meaning that you can see the components side by side and identify Trends uh just like before you can just filter it down to like, in this case vinyl and the streaming formats, uh which uh will it's basically just zooming in on two of the components, uh and uh you can get back to it, just the original chart, just by uh emptying the filter.

E

This is what my wife call when the elephants kiss, so you got back to the original chart and also you can just switch it to a line chart again either with the original syntax, so that you can compare the trends in this way or with the config line, which will just enable you to to build a line chart a lot easier. So we we addressed the first issue about the generic logic, but the one directional use case for that. We built this new extension that we released earlier. This July called IPI visual story.

E

This is separately. This should be installed separately. It contains things like stories, slides and steps, and basically it enables you to build interactive animated presentations with the in Jupiter notebook. So, basically, here we just set the style and add some more data into it.

E

I'm just trying to be quick here, you create, you, add the first slide and you add a step, and this step is basically what a chart that you had in I5 is a story, and then you set it up, and you add it at this slide that you put the step into to your story. You do that again, like this time, we're going to split it, it's a little really simple step and so on and so on.

E

So you can construct their story like Slide by slide, you can have slides that contain multiple steps like in this case. We have slide number three here and then we add one step here and another here. These steps again are these very flexible charts of of ipivizu, and you were going to end up with something like this. If you run this cell, this is uh fully interactive, animated uh data story. Here it also has the tooltip. If you move your mouse over, it will show you the data element there.

E

It actually shows um the uh uh the budget uh that the the U.S federal budget spent on r d uh between 1955 and 220, and you can move between the slides either with you cursor uh or by click of a button or with with the buttons on your keyboard or even with the clicker. As I said, you can once again see the components side by side.

E

You can add, in this case the defense budget, which absolutely changed like giving you context of how much uh the U.S spent on everything else uh throughout the years or you can just yeah. You can once again see the the share or you can just uh you know, zoom in on the beginning and the end of the period uh group, the components of this Tech chart and then rearrange them so that you can see how these components changed over time. These are just simple examples of of using ipivizu an IPI visual story.

E

You can also export this thing to an interactive HTML file that you can share share with your peers. It's going to include the data it's going to include basically the chart that everybody can run in their browser independently from The Notebook. We have the whole thing open source you can find it on GitHub.

E

It's called that this is when I type is a story and uh yeah I really don't want to take more of your time without letting you ask me some questions, so thank you very much for your time and I'll be I'll, be glad to hear uh what you think about it.

A

Thank you very much for sharing one thing. I want to check before we discuss. Do you have links to those projects, or is it an IPI visual org, where I can find it.

E

um You mean the ones that I just showed you with the music data and everything like that.

A

Anything that you you can share. Oh thank you. That's.

E

Wonderful, here's a repository link and here's an example in an HTML format actually from the repo you'll, get to some examples. Some uh the the presets that we built for ipivizu, uh actually a step-by-step tutorial. We created for the thing uh and we also have a slack Channel. We I mean I personally thrive on welcoming everyone who joins and answering every question that might come up. uh Obviously this is still very fresh and we know there's a lot of things. There are a lot of things we could do to to make it.

E

You know easier to use for everyone.

A

Yeah, that's awesome too. um I definitely have questions. I can ask but I think other people might so take it away. Anyone who has the question.

B

uh I'll uh I was really impressed with the animations. They were very clean.

B

um I actually went into it being like Oh I'm gonna hate this, because I don't like a lot of color movement, because it I do get a little bit overwhelmed with those experiences, but I was really impressed with the transitions between the slides being as graceful as they are. What what's? What are you guys bringing together here? Like you, it's a really cool story, but you know like from a technology side like the static plots, look as good as other ones. That I've seen is the animation and kind of the uh in between steps.

B

What y'all are bringing here um I mean it was a good story, I'm still trying to I'm just trying to figure out. You know where you guys are focusing your interests.

E

Sure uh so, basically, this whole concept comes from my co-founder who's, a ux designer and who was really frustrated by the fact that he has to learn coding to make animated uh really, like you know, bespoke visualizations, and uh he started to think about how how this can be, like uh short, how how what kind of a shortcut can be built here, and he figured out that in order to build a system that enables just basically everybody to to better interact with data, it has to be like a single logic based on a single engine that can generate all the charts so that it can be animated, because animation is key in keeping uh people's attention and just keeping the context, and so he went into a Langston research to identify the common characteristics of all the different chart types and also to try and come up with rules for uh the animations between these these charts.

E

So there's there's like yeah literally years of research, behind those clean transitions that you supported. So thanks, I will tell him you appreciate it that he'll be very happy about it um and uh yeah. We we are actually like. Our vision is to build a graphical user interface for data, a new sort of language for data, uh and uh that's a that's a huge task. To be honest, uh we are happy that we have the initial version of the engine that can generate all these charts and the animated Transitions, and we can we.

E

We were able to start sharing this experience with others and it seemed to us the data scientists are the more most well like the best fit uh as as early adopters for for such a thing. So that's that's how we ended up here.

B

Very cool, thank you.

E

For the question.

A

I well, what do I I'll go on if someone else is going to start talking.

B

I got a question about stories: yeah.

A

B

Mean I I, so you know I lit like one of the things I've noticed over the years is a lot of people have put effort into plotting tools right um because, like data scientists are often dealing with more data than other people make decisions on. So we focus on like the form of the plot, but then like now, when I'm delivering reports to people, you know I'll have three and four plots in there and if I do a selection on an axis, it effectively changes the whole text.

B

You know um and the way that that story goes uh so personally, I, don't see the stories completely propelled by charts, does like text and live rendering of text play into this sort of the you know: Brett Victor uh Eventing on principle, kind of Concepts.

E

um I, we we don't have anything against the textual descriptions. I mean uh that that's absolutely like whatever helps people to make sense of what they're. Looking at uh the story that they're being told the insights that somebody else found uh it's just that ours, our approach and our our specific, uh like benefits or or capabilities of our system of our tool is lies within visualization we'd.

E

Be happy to integrate this into uh like there's this startup that Tableau acquired last year called narrative science and they already released it within their product that basically uh tries to translate. What's on the chart to like a couple of sentences uh to describe to help people understand what they're looking at, we are not a goal. It's it's like we're, trying to find the the most effective, most efficient forms of communicating uh insights and to help people interact with data.

B

The storytelling is compelling I think you've got something there good. Luckily,.

E

Thank you Tony. Thank you. If, if you have some time to play with it- and you know we are all, we are all very happy to receive a candid uh and and harsh feedback even about what we should improve on, we know it's it's far from perfect and we don't want to act like it's ready or anything.

A

Can I ask a kind of clarifying question? So are you sure so one of one of the things that I think I noticed right is what seems to be happening. To me is that the different right, outputs of whatever the chart is like at its different states, are then used as the key frames for the animation, but um within that, is it always kind of going like the same way that elsewhere in a notebook that the order that you run the cells would impact it like?

A

If is it going to try and just between between, like whatever the last thing I run was with that like say, I skipped the cell directly before or does it like? I was kind of curious how it collected those because it since we were going in order I felt like we were seeing it very smoothly, but I was wondering how it knew how it interacted with the other cells, since each keyframe seems to be being made in a different cell.

E

Yes, that was uh with the original iPad, the zoo, with the ipadvisor stories of the presentation part uh you had all the these states or keyframes as you just called it uh in one cell. You can actually have it in multiple cells and then there's an order that you said. Actually it's I mean I I routinely make stories like this and I when I want to rearrange something I. Just you know copy and paste the code somewhere else within the story and also you might have noticed.

E

We have the go to first and go to last button just next and first so there's a chance for you to to uh go not in the preset order, necessarily also since it our tool, works well with ipy widgets and also it can be implemented in mode and streamlit and and other like interactive Frameworks. uh It's it doesn't necessarily have to be only a story with a linear path that you take the audience through, but you can build some sort of interactive explorers.

E

We also have some examples for that, uh but yeah it takes a lot of time uh to to build such a thing.

F

E

To running an artistic startup which I wouldn't suggest for anyone to do it's, it's really uh yeah. It takes a lot on it's. It takes its toll. I would just say: oh you're, on mute, yeah.

A

No I was just nodding.

A

um I always like to hear the the candid advice with that. Thanks yeah.

B

A

That makes sense. I I just find it very interesting to see how that's going out I if I can ask to, and people can either stop me from going this route or interrupt for a different question. I also just think it's really interesting right, like you were showing us in many cases. Example data and kind of just trying to Showcase right how the animations worked like the way that you would invoke those animations, but I also find something really interesting right like when you were showing the defense budget data.

A

There was the moment where you're like oh we're, going to compare this using right. The defense budget in the U.S is kind of the 100 so that you can compare everything directly to that and so I I believe that's what you're getting out with the storytelling right is that you can show things over time or you can show the different views, as you said at the beginning, but I almost wonder with that too right. There's a lot of discussion I've run into on the on the periphery I'm, not too involved around like right. I!

A

Wonder if there's like a potential for.

A

Sorry I'm trying to figure out how I'm going to say this.

E

A

Is there ever a case where you would say it's really easy to misrepresent the data? It's kind of difficult, because I know when you're making the framework that you can't necessarily control right control, all the content that people do with it, but I'd wonder if you've run into anything like that yeah.

E

um You you cannot control how people are using, especially an open source.

B

Frame, you can really and.

E

If you don't intend to do anything like that, lying with with data and I mean there's a like a great book from Alberto Cairo just about this topic and there's like plenty of examples, it's fairly easy to do that in many cases people are you know just playing with the access ranges to to show something. That is a tiny change to to enlarge that. This is just a tiny example of of such a manipulation. There is no way to do that.

E

I think uh like enabling people to to better uh understand what they're looking at and to interact with it not just take something for granted that they just saw, but also make it really easy for them to switch from storytelling to like okay.

E

Let me just take a closer look and play with it myself uh and and keep making the bar to lower and lower, bringing the bar lower and lower that that just enable people to to make their own analysis or- or just you know, look at such stories with with some criticism and being able to you know just edit. The whole thing that they got is can be, can Empower or can be some sort of a defense mechanism.

E

To this thing, I I think actually jrv wants to ask something: I think because it was he turned on his camera and he even unmuted himself for a time. But that's just a guess.

F

Hey yeah yeah, um hello, uh we we talked, um we talked to the best because I was actually building a template with uh visu to aggregate it to add it into Nas. Catalog and I was just asking myself where I was playing about hey. How do you feel about those charts and, and how do you you know, because you don't expressively say this is a horizontal bar chart or this is a um a line chart or in the configuration you you said it in a presentation.

F

So what's your take on this and how are you seeing the this this aspect of the configuration about like mentioning which kind of charts you want to see: I'm, not sure I, really I'm, really clear, but tell.

E

Me no you're, absolutely you're, absolutely clear, and this is what we try to address with the chart presets. So.

F

Because this was the only kind of roadblocks that I faced when I was building the templates um and and I just wanted to have.

E

Your thoughts about it, yeah yeah: when did you do that I'm sorry, I, just.

F

I was actually building them yesterday. um You know we talked a few weeks ago about adding them to to to the nas uh template. Catalog and I was just playing around because I saw you were presenting today, so I'd say: oh I'm gonna put some time in there. Yeah.

E

So um let me really quickly uh show you something. uh So if you go uh to ipivision.com and check the examples, then the first ones will be preset charts that are on the on the on the side, and these are all all uh like constructed in the way that you I think you were looking for. So here's like I, clicked on the percentage column, charts thumbnail, and this is uh so how you should modify your code. It's not just config, but config dot, percentage, column and you add the series this way.

E

So it's more uh straightforward and you know uh straight away, what's the type of chart that that you're building, so we have actually here uh 40 plus chart types that you can use as.

F

Yeah, you have an amazing Library.

E

And it's it's a lot easier now to to build uh these kind of charts if you click on any of the thumbnails in the preset Gallery uh you'll end up at uh an example uh code that that I'm at least I hope uh make it really clear uh what what how it works. Authority.

F

E

Parameters of the chart type.

F

So basically it will not be. uh um You will not declare the chart, as uh as you will not really name it, but you will enable the the way the chart is displayed through this configuration that you showed right.

E

You you actually name it so like we used now uh percentage column, that's the name and you're building a First Column chart or.

F

You're: okay, okay, scatter plot.

E

Config dot scatter and then it has uh certain parameters uh that that are necessary to build a scatter plot. So we we are getting closer uh to to to that. uh Maybe just you know give it some if, if you played with it, if you have some more time and check out the presets I'll be happy to continue our discussion uh to to understand. If there's still something missing be unclear for me at this point,.

F

No, no, that's! That's all good yeah. What you're looking I would be happy to to to no I will I will I will let you know. Thank you so much for the for the answer very clear. Thank.

E

You it's not so much in build. Thank you thanks for the question thanks.

B

um Peter I don't know if you're familiar with uh Hollow Vis project.

E

um I I've heard of them briefly I heard about it. It's.

B

Not too much work to build uh visualization back end for hollow views. um They have an API, an extension API, but the nice thing about Hollow viewers and Hollow viz is that it sort of generalizes your data ingest and then allows you to you know: take a pandas data frame or a desk data frame, or you know mode and data frame and like move that into your visualization real fast.

B

But there's like to me, this looks like Altair, with like a specification or like an Altair Vega, with a specification for animations in the middle um and when you use when you build an HV plot or a hollow views back like visualization back end, it actually turns your pandas data frame into a fluent API for visualization too. So you'd actually get an alternative API for writing and composing your visualizations. The way that you are so you can say something like df.plot group, this thing filter.

B

This thing show this thing and then you know you can feed that into your system and have these really beautiful, visualizations coming.

E

Out um yeah, actually, now now that you tell me it's just, they have so many different names. Honestly uh I mean uh the the whole of these guys. They have the panel. They have other things. So actually, there's uh This Thread that I just copied into the chat where uh Mark who's I think works at Hollow is, uh did an example of of using IPI visual uh within their their framework, I'm, not entirely sure, because I have to come. uh I have to do it.

E

I have to tell you that I'm, not the technical founder here, so I'm not entirely clear about what the they do, but it seems that we work nicely with their panel thingy I'm, not sure. If what you just said is any any way different than this or.

B

Hey it's great that Mark is on it. Mark is a really good flight promoter for all of you. So if he's playing around with this stuff he's going to build he's going to build some things with it, so you're in the right place.

E

Great thanks, Tony.

E

Yeah we have some more questions: Isabella I.

A

Was gonna check that anyone else said I do but I also know when we're running um watching the time that I can have longer conversation, because that was really awesome. So I'm gonna give people a few seconds to say more. But oh, we have one more question: yeah.

E

Tony yeah does this work in lab or just classic? um It should be working in lab. uh We, we are double checking it these days, uh whether we have all the functionalities.

E

If you run into anything, just give us a shout out, uh we are actually, as I said, very very approachable. uh I I was thinking like maybe a question from my side uh is: would it be okay to open a separate, like thread in in the uh discourse uh Forum about ipivision iPad with the story, so any suggestions and questions and issues that you might run into uh would be very welcome here there. If that's, okay, for you.

A

Yeah, that's absolutely okay, I think we have in fact a lot of things on the discourse that are like. Oh I, want to. You know here's my point of contact for this project.

E

um Thanks yeah.

A

No, that's uh that's I, don't know, I'm excited for this. At least this is really interesting. Although I have a bajillion accessibility questions just so you know that's why I was like I'm stopping myself here, because that is a hard problem just to say for you. That is not just you. You we're not the only one um who might be running into territory list, but here I'm gonna send you one link, though just so. You have a point of reference I'm going to send you chart ability um which.

B

Is yeah well I'm, going to send a second link uh which is Apple uh Apple's Design Group just released um some guidelines on charting data yesterday. That seems to be pretty nice and influenced by some of the work that Isabella is gonna I.

A

Was gonna, say, I'm pretty sure, there's some overlap on the team writing those yeah, um but these are.

B

Good resources.

A

Because it is a really hard problem, but I think that there is actually a potential for this to be. You know kind of an accessibility Boon in cases of right, uh like cognitive accessibility, but it's going to give us some challenges in other places. So.

E

Yeah um the I mean vvr, like this co-founder of mine, who came up with the original concept of uh visu. He's he's not colorblind, but he has issues with colors. So we are all I mean I I know it's not the same. There are different levels of accessibility challenges, but uh because so we we have this sort of personal approach to this, that we do want to think. It's really just a matter of resources uh on on our side and and then I can refer back to the early stage.

E

Startup thing that I mentioned before now.

A

No yeah, that's why I just wanted to link it. I know, personally from the case and other projects that it's not a small task. It was not meant to be a call out. Sorry, but.

B

I think in some way um this actually is an accessibility tool for people who we consider themselves abled, because, most often, when we're observing color, we really observe the changes in the color and the fact that y'all are bringing this change of color in between the graphs and somebody's eye can follow.

B

It um I think it creates new Pathways for better cognition, also new Pathways for obfuscation, but in some way or another, like this connection of color through the story, as things are moving and that continuity is going to be a really power like it's a really old design concept and, like you guys, are like sort of rediscovering some old ideas and making them new again, which is pretty cool.

E

Thanks I I I, thank you, I I, I didn't really know what to expect from this call, but I'm so happy that you appreciate what you do so much and then yeah thanks again and if you have the time to again to to just play with it and let us know how to improve that would be. That would be absolutely absolutely awesome uh and if any of you want to continue uh chatting here's the link to join our slack and- and this is my my email address uh so just you know reach out to me.

E

If you have any further questions comments, anything.

A

Can I ask, are these things that you want um stored after this call and the notes or just leave them in the chat.

E

Yeah, absolutely okay, please.

A

And thank you so much belated welcome to the I think it was in chat when you were talking, but yeah I'm really happy that you decided to join.

E

So, thank you. I I really feel well. Thank you.

A

Final Call for questions on this. Now, though, definitely not the final call for iPod the zoo questions in general.

B

Have you presented this at a conference yet.

E

Yes, I was at europython and I was at uh padeta Lithuania and I'm also invited to present this at uh by data New York. uh This November.

B

How do the rounds of Applause feel when you show this stuff.

E

B

E

So I'll see you there then.

A

uh That's good for me, then. If people don't have more questions, I can open it up. I know. At the beginning there was some mention of: do we have time for other questions that doesn't still have to be the case, but I'm happy to spend some time on that. If people want.

D

Oh yeah, so um are you guys some Uzi and um I've been working well I've been using Jupiter lab. Let me put it that way. For a long time, uh inside a um and um lately I was like really I got relocated to Singapore and I figured. There are a lot of firms here, trying to adopt Jupiter lab into their internal data. Science tool and I was wondering uh first question: do we have a more or less a like a Jupiter Community branch in like Asia region or Apec region.

A

Do you mean, in the sense of like having a call like this, or do you mean more like broadly like? Do we have a community.

D

Oh yeah, do we have a community-ish because we I was thinking if we could be setting up a like Apec region, Community for Jupiter lab and help uh the Enterprise to adopt Jupiter lab better into their ecosystem.

B

I think you're probably going to find some more luck in the pi data route. um Like a lot of Jupiter people are Pi data people, um uh yeah I'll, send you so Pi data. um So we've got a couple uh organizations that help support our open source communities.

B

um Two of them are Pi data and num Focus. um So Pi data is a global community of meetups around like the Scientific Python data science, space um there's over a hundred different Meetup communities, um so we're more likely to find uh Jupiter use. Jupiter users, business users, industry users at.

E

B

Events um there have been a few small communities that tried to be Jupiter specific, but I, don't think any of them have caught on in a sustainable way. But.

A

B

I would say.

A

We may not be running into them. They like that. Actually I would say that we may just have a kind of awareness Gap there I don't know. Tony usually is very aware of all the happenings, but I would say in the past, when I've tried to schedule events to be more inclusive with that time zone at the very least, I've never found the groups that are there. If they're there.

E

um That doesn't mean they're.

A

Not it just means I feel like we haven't connected across that, yet which is honestly pretty possible that they exist and we just haven't overlapped at this point. Oh.

F

I think yeah sure yeah.

A

C

um One thing is that we do a set of um calls for proposals to do community workshops that we fund as a project and I, think you would find a very receptive audience to try to fund a workshop and try to kick-start a a group that does meet regularly and also if we can find ways of so. We have a lot of regular calls. You've, probably seen the Jupiter community calendar.

C

In addition to this monthly call, we have lots of projects that have weekly calls or bi-weekly calls, and if you find that there are some of those calls you're interested in and you're, basically locked out for time reasons.

C

I I can't speak for every project, but I do know the ones that I'm involved in would all really strongly consider alternating the times and figuring out how to be accommodating it's just because we haven't had points of contact that we already know. It's been hard to sort of seed that ground, but if, um if if we knew that, oh doing this enables a group of people to come, we would really welcome that. So um yeah.

D

Yeah so uh just as you mentioned, the call for proposals I was actually reading through it and trying to submit my proposal instead of Google The Google form assessment has been closed already, which is like a bit confused and don't know who I should be contacting with.

C

No, no, that that it that it's true we do this I think once a year basically, and we found a bunch of projects and we've probably done this round of funding.

C

So the next hormone I, don't exactly know when it is, but I imagine probably will start up again in a in a few months uh when the New Year arrives, gotcha yeah- and these are not meant to be like this- isn't meant to be a replacement for like say, Pi data like Tony said this is more like if you have a a targeted thing that you want to do, or you want to kick-start some project or you want to like, have a Sprint for something or you know like they're meant to be like sort of focused workshops, and um we we usually get funding from a couple of long-time corporate contributors to the project.

C

Who will just say we're allocating these funds to you to do community-based events like that and um so yeah, so the next time it comes, we'll, definitely announce it on this call and and lots of other channels like our blog Etc. But beyond that, before that happens, I'm saying like if, for example, you're interested in Jupiter lab or, for example, you're interested in Jupiter server, um we will definitely be open to scheduling at least some of those costs to be a friendlier times.

D

Gotcha so yeah, so uh that's that's one thing. The other sentence like I wonder if there would be uh any sort of way for me to get to know like how Jupiter lab is like deeply integrated with um Enterprise IIT infrastructures ish, because I do I, do no good amount of them back in Asia, but I would definitely be more interested in how people are like leveraging that in North America or in Europe.

D

Do you think there is any way we could do that.

C

I'm, not sure I understand. Could you refresh.

D

Yeah so basically I wonder if there's any way for uh for people to like know how Jupiter lab has been integrated with Enterprise infrastructure. Deeply like talking about like how do you access like to your like production data, with the Jupiter lab, how to protect the the um the the the um the environment safety stuff like that.

A

All right, are you asking or can I check like case studies or tutorials like? Are you looking.

D

For pupils I think I'm, looking for like case studies right now, uh because, as as also at the same time, I do see a good amount of my friends are looking for how to like integrate YouTube Jupiter lab into their internal tool chains.

B

D

So a lot like a lot of.

B

A lot of that is uh happens at sort of a different scope than this meeting. um A lot I think you'll have some good luck. Looking at the Jupiter, Hub community um and you'll find a lot of different use cases and Integrations of multi-user systems kind of in the Jupiter Hub level, um and they also have a community meeting. That's probably going to be more um Asia, Pacific friendly, so uh yeah. That would be a suggestion. I'd make.

D

Thank you. uh Do you know, how is there any like a link link there? Any like URL could be linking me to that site. Ish.

B

I think there's a I think Jupiter Hub, slash team compass um on GitHub might do the I'm not totally sure.

D

A

Are you looking for Tony.

B

Something to point to Jupiter Hub, um because Jupiter Hub is like really at the systems level and, like you know, I think Netflix had an article a while ago. uh That was good about how they integrated some of their data. um There's some case studies here and there, but it seems mostly Jupiter Hub hooks everybody in.

D

Culture, so Jupiter Hub is pretty much like the repo right now. People are like putting putting stuff together to see how things are getting architect at a multi-user level. Right, I,.

A

Don't think it's maybe exclusively, but it's definitely the hub for it. haha Okay,.

D

Gotcha yeah no I was because like when I was architecting around I was like I was thinking totally I was thinking another way of doing it, but thanks thanks for the redirection.

A

Yeah can I also add circling back to kind of your first question. With this community call in particular, we I've tried to kind of rotate, sometimes it's a little hard as a single host to dodge other meetings that I have and uh when I'm a good host, for example, I'm, not a great host at 1am, which is like I, know an alternate time that a lot of people at 1am.

A

My time, I should say, is like a good time for people with this um I've asked for people like if people want to host before or co-host or anything that is open. So I'm also just opening that up to you, if you're interested in having similar like cross community sharing calls, we can talk more or you can come to the I think I shared the repo. If not that's fine too I just want you to know.

A

This is not something that is like I, don't know controlled by someone like if we're interested in having other community calls rotating times. For this, too I would be very happy to do that figure. It out.

D

Yes, thanks. That's that's pretty much like all questions. I have that I know. Thank.

A

You so much for asking them. Those are great questions, I, actually love. When people come to this call with questions. We usually have really unusual discussions.

A

um Yeah anyone else have something they want to add. I know we are over time. People can uh go whenever I just want to make I actually have to go, but I want to make sure nobody's being cut off cool. Thank you. So much all for your time today. I hope to see some of you. If not all of you next month, uh end of October. We will be here again, so thank you for another great month of calls and good luck with everything in the meantime see ya.