Transcript#

This transcript was generated automatically and may contain errors.

Hello friends and welcome to episode 19 of the Shiny Developer Series. My name is Eric Nance and I am so happy that you joined me for this episode. This one's an exciting one because you've heard me say in previous episodes that if you're getting into Shiny and you're trying to level up your skill set, this is really one of the best times to be doing it. You've heard some excellent guests in our previous episodes that have been authoring some great materials on leveling up your Shiny skills and the guest today is on that same thread so to speak and it is none other than RStudio's chief scientist Hadley Wickham, the author of the soon-to-be-published Mastering Shiny. In this episode you're going to hear us talk about the origins of Shiny from his perspective and get some of his advice for just starting out with Shiny and for those that are already familiar with Shiny we're going to talk about some great lessons with debugging, how to level up your skills and best practices for seeking help and contributing to the Shiny community. Well let's not wait any longer. It is a real thrill to start share with you our interview with Hadley Wickham.

Well hello friends and welcome to another exciting episode of the Shiny Developer Series and I am absolutely thrilled to welcome for my guest of this episode chief scientist at RStudio, architect of the tidyverse and many other accolades that you probably know about. I'd like to welcome Hadley Wickham to the Shiny Developer Series. So Hadley, thank you so much for joining me today. Yeah thanks for having me Eric. Absolutely. So certainly many in the community I alluded to it earlier know you best of your contributions to the tidyverse and many contributions before that but not many might know about your involvement with the early days of designing Shiny.

Origins of Shiny

So first maybe in your own words where do you why do you think Shiny is so important in the ecosystem and then maybe you could elaborate a bit on how you were involved with the early days of Shiny development. Yeah so I remember I don't remember exactly where this fit in the sequence of kind of Shiny development but I remember Joe and I were in San Francisco together and we visited a bunch of you know kind of like big internet companies like you know like Airbnb and Pinterest and places like that. And one thing that came up kind of again and again with our discussions with data scientists is how like the struggles of like turning what the data scientists were kind of working on into like dashboards that other people in the company could use. And you know a very common pattern you know at the time I think it still exists today is that the data scientists would like do the thing and then they would turn it over to the web development team to make the dashboard. And then you know there's a lot of back and forth explaining what everything means and how it's supposed to work and you know it takes a long time to get on the web development team's schedule and you know like six months later you finally get the dashboard and then by the time it's up like it's basically out of date already. And this is kind of you know coming up a very common struggle a lot amongst lots of people and I think that really that kind of really drove the need for Shiny. Like how could we give our developers the ability to create you know compelling web applications that they can then use to surface what they've been working on.

Great yeah that's great to hear that back story and certainly I mean for me personally it's been a huge revolution in my workflow and it's enabled so many opportunities that as you alluded to before that we would always have to hand off these different stages of the analysis to another team and there might be you know gaps in timing gaps and understanding but now with the data scientists or statistician involved in that end-to-end process of developing these applications it's it's been a huge win in collaboration across multiple industries and you even alluded to the book which we'll get to but my industry of life science is one of the biggest customers of Shiny so certainly we even had Joe Cheng give a keynote at our first R&Pharma conference and he was genuinely surprised by all the workflows we were fitting Shiny in so we are very much appreciative of that. We've also had I know I've had like at least two people have like thanked Joe and said like you've saved me like several hundred thousand dollars. Oh absolutely. We were like we were like looking at hiring a company to do this and I just started playing around with Shiny and then like over a weekend I like created the app you know the basic app that we needed. Yeah it's truly amazing yeah both the time and the money and frankly the collaboration enhancements are just immeasurable in my book so speaking of yeah so speaking of books you know I was of course extremely excited when I heard that you were authoring Mastering Shiny so maybe you could tell us a bit about the backstory and what inspired you to author the book.

Writing Mastering Shiny

Yeah so I guess Joe is one of the victims of my book writing advice which is like writing a book is I think it's quite simple like all you need to do to write a book is like write for like an hour or two every morning and that's like that's the secret to how I've written so many books like that like that's all you need to do. Interesting. It turns out that that advice like I've given it to a number of people and no one has managed to use it successfully because I think like the writing for an hour every you know writing for an hour or two every morning is actually the the hard part and it's like this is something like clearly that I've internalized that allows me to do that which is which I don't know how to put into words so Joe had actually been working on Mastering Shiny for I don't know like a year or two maybe and I was just like really just having a really hard time of it and I just thought like you know I'd love to help him out love to learn more about Shiny and how it all works and I you know had a kind of a an open slot in my book writing schedule and thought it'd be kind of fun to tackle it. Wow that's a that's a certainly an interesting set of circumstances but it's also maybe a unique time for you where you're writing about a concept or a package here or a workflow that certainly like we alluded to you've been involved in kind of the early genesis of it but it sounds like you were learning more about it as you were writing the book and that's pretty fascinating as well it's a great way to learn.

The other thing I really like about it is like compared to alpha data science like if there's any like when I'm writing about something in alpha data science and it just feels like like annoying or hard to use I feel like I have to stop writing and then go and fix that little thing like it's someone else's responsibility so I can just kind of write about it and say you know this is like a little bit of a rough edge it's not a big deal just work around it by doing this which was you know I've also filed a bunch of issues and pull requests to Shiny but just nice to have that to kind of write it it is nice to like write about someone else's work and just that that distance is I don't know I found that really really enjoyable in the writing process. Yeah I did notice some activity in the issue trackers for not only Shiny but also some of the extension packages like Shiny test and others so I knew you were definitely learning and seeing some issues along the way so that's great to have that collaborative development on multiple fronts.

And you alluded to early obviously you know your previous efforts in authoring like of course R for Data Science and Advanced R and as I've been reading Mastering Shiny in the early web form I have seen themes at least from my opinion where you're taking learnings or concepts from like how an R for Data Science is worrying about the analytical workflow the data science pipeline and then from Advanced R it is certainly coming from an engineering you know going inside R how it works perspective so how do you think something like Shiny blends those two different frameworks together in your opinion?

Yeah I mean Shiny I think I mean I kind of think about like there's at least a I sort of bucketed in my mind a little bit differently to you like I think Advanced R is being about like programming okay like how do you like write like how does R work as a programming language and how do you do stuff and then you've got like R packages which is more about like software engineering like how do you write big like big bodies of code that like work and maintainable and can kind of grow over a longer period of time sure and then you've kind of got the data science side and like Shiny kind of touches on all of them like you're using Shiny you know primarily to do data analysis you've got to learn about this new like programming framework of reactivity and then you've also got like because Shiny has been so incredibly successful people are like writing bigger and bigger apps yes and then so you also need to learn like a little bit of software engineering so that as your app gets bigger it hopefully like the pace of development stays constant hopefully you know maybe if you're lucky it actually increases as you get better at it rather than like decreasing over time which I think is the big like if you're if if if you're like working on a coding project and it just gets like slower and slower and slower to do anything over time I think that's a sign that there's like something wrong with your kind of software engineering process and that that's kind of like one of the things we see a lot in um with Shiny apps like people people make these incredible apps like they've all been surprised by like how far that people have been able to go to make these really useful apps and then we look at the code and we're like oh my god like what are you what are you doing that's which is not you know it's not like the fault of the like it's like a testament to like you don't know that much about software engineering you've created this incredibly you know incredibly useful thing but it's really difficult for us to kind of look at and say like this is what you need to improve because it's like you know you need to like go back to kind of step one and so that that's hopefully like one of the goals of this book is to give people those skills so that you can start off with a simple app and you know it gets more and more featureful and complicated over time but it doesn't become this kind of like unmaintainable nightmare where like you you don't understand like whether you know your change you know with the one reactive ends up firing 50 other things what's going wrong.

so that you can start off with a simple app and you know it gets more and more featureful and complicated over time but it doesn't become this kind of like unmaintainable nightmare where like you you don't understand like whether you know your change you know with the one reactive ends up firing 50 other things what's going wrong.

Yeah i can uh i'm i'm certainly laughing from past experience um so i was one of the early adopters of Shiny and i my first i would say major app it was exactly the trap that you you just alluded to i built all these different things together and this was even before modules became a thing which i'm sure i'll hit on a little bit and trying to network all that together is just a nightmare but more importantly i cannot instill like what my philosophy was to other collaborators or frankly even the future me so it's certainly a lot of these principles that don't come natural right away but once you once you adhere to them it's a better product and it's a better experience for sure so we we can all sympathize with those things yeah and like i don't want to like you don't want to feel anyone to feel bad about like the quality of the card because the only way you can write good card is to write like a lot of bad code and that's exactly you know what i what i have done oh yeah i remember you said that in one of your previous presentations exactly yep hopefully we can just skip you a few steps so you get into good code without going quite down quite so many dead ends yeah absolutely.

Getting started with Shiny

So yeah let's maybe dive into a little bit of the structure of the book and one theme i want to touch on first is you do a great job of talking about you know what are the best ways to get started with Shiny certainly there are resources about it but it's great to kind of see a kind of recipe driven approach to how you create your first app but my my question to you is as somebody who's getting involved with Shiny maybe for the first time or have basic familiarity with it what are some of the expectations you think users should have before they embark on this journey as compared to maybe getting into say the tidyverse analytical workflow or things like that in your opinion i think probably like the two things to me the two biggest differences are probably um like you're gonna have to learn reactivity which is like a fundamentally different programming model and like it absolutely takes a while to get your head around yes like i still you know things still surprise me and i have to like do little experiments or just ask show why it works that way sure um and then i think the other part of it which is another area where you can just like there's just so much to learn is like user interface development like how do you actually make an app for people but other people who are not experts can use effectively right and you know i don't get into that into the book at all because that's just this huge huge topic really really important and but but thinking about that and learning a little bit about that i think helps really helps make apps more effective.

Yeah it's it's just interesting to see the different mindsets that come into play here and yeah there are there are definitely what's great about this time and like learning Shiny is that there's a lots of great resources now that are hitting different themes even in more detail if you need to dive into those worlds so actually respect the user interfaces david grandgen is authoring the outstanding shiny ui so that's a great place to go if you need to get in the nuts and bolts of that alone but when i first started Shiny the best the only resource i really had was the google mailing list where joe and others are frequent and that's how we were trying to learn so it's a much different time now and um another part of the book that i was really excited to see as i've been adopting more of the tidyverse related packages in my a lot of my work but including my Shiny app work is how we can leverage tidy evaluation in developing your back end Shiny code and you touch on a couple great concepts that i didn't really articulate well in my head before is that within tidy evaluation you have these constructs of n for e and v variables and then these more of

the data variables maybe you could elaborate a bit more to our audience what that means and how that fits into like using tidy evaluation with shiny yeah i mean so tidy evaluation i think has had a lot of ups and downs and it's history i think like two years ago maybe someone posted on our studio community saying like is tidy evaluation destroying the tidyverse oh i saw that

i think part of the challenge was like the it was so obvious to us like the theory behind tidy evaluation was kind of correct and gave us a lot of power but it was it took it's taken us a very long time to kind of crisply explain like what do you need to know to actually learn it and to use it sure i think what it's come down to now i think what kind of captures why it is so difficult at first is this idea like we use the word variable to mean two different things and until you kind of disentangle those in your head it's really difficult to use tidy evaluation because you don't have quite the right like hooks in your brain to hang things off them and it's basically the idea that like in r you've got like this we use variable to mean two different things we use like a a data frame variable that's like you know in the m in the in the iris in the iris data frame there's a variable called species and we also use it to talk about kind of like objects that exist environments in r like the the iris data frame is a variable in the the data sets package right

right and and the the thing that i think makes tidy evaluation kind of challenging is now you've got this like extraction you've got the name of a data frame variable stored in an environment variable and first of all like that like until you've like separated the two types of variable like that seems doesn't even make sense right you've got a variable that's stored in another variable like what does that what does that even mean and it's that but it's that indirection like that that that's really at the heart of tidy evaluation that that being able to capture like this is the name of a variable a data variable i'm going to put it in an environment variable and later on i'm going to use it in the context where it's going to be a data variable yes

um and so i think i mean my hope is like that that now like once you get that kind of idea in your head and and then the problem of the you know the great thing about tidy evaluation like as a from a user perspective is that you can use data variables and environment variables interchangeably and that's what you know that's like when you're subsetting a data frame you can just say species equals you don't have to say iris to all science right right but it's that ambiguity that then makes it hard to program with and the basic idea is like once you've identified what's the data variable what's the environment variable you can now use these these pronouns called dot data and dot n to kind of be explicit about where things are coming from and then you know after that i think things just kind of relatively naturally fall out at least at least for the case of using it in shiny apps which is a you know slightly simpler case than all of the general possible cases you might want to use it right and it reminds me back when i was taking one of your workshops at rstudio conf it was about a couple years ago when you're doing an advanced tidyverse kind of workshop and when you were talking about tidy evaluation at that time i didn't it didn't really click with me this concept of the different types of variables but now after reading this chapter it it once you start there like if you start there then you're able to decompose it successfully it becomes much easier to harness than maybe when i first tried it a couple years ago so i know you like you alluded to you've learned a lot along the way you and others on the team but i think it really shows in how you've authored the material here and i think it's a great win for shiny authors to leverage these concepts effectively especially if you have like back-end code that's like doing data processing data manipulation and you have functions that are already doing this kind of thing you don't have to like mentally shift to a different paradigm if you want to use it again so like i said i was that was extremely excited for me to see that

yeah it's great yeah yeah i mean am i an expert at it certainly not yet but i'm i'm known enough to be dangerous as they say so it's been exciting to learn about that and um the other concept you alluded to that i think is the most fundamental in shiny in my humble opinion of course is reactivity it's really what distinguishes it from a regular r kind of analytical workflow here and so what's your opinion you kind of alluded to it a little bit on some of the more challenging aspects of reactivity that you can the users need to grasp across the different maybe spectrums of experience that someone might have with shiny yeah i think the

Understanding reactivity

i think reactivity is kind of challenging because like when you when you start using it like very simply and very initially it's you know it's like magic but it just works and you don't have to like think you don't build up like a mental model of what's happening just your like outputs update automatically as your inputs change and if you've never really thought about it like that just it just feels like that's how like that's just obviously how the world should be like and how the world should work but like as your app starts to get more complicated you need to start to kind of make that reactivity maybe like a little bit conditional you only want to update this thing if that thing is true it's very easy like you to kind of get yourself into the state where you have no idea what's going wrong and it's just just magic and you're just like in a randomly changing things in the hope that it'll that'll um fix your app and i think at that point like then you really have to sit down and like say well now i'm going to spend a little time like really trying to understand the theory and that's when i think like starting to think about like the reactive graph and the arrows connect you know that that that's like picture that's what i picture in my head is like when i'm looking i'm going to show this how do all these things connect together right and right like the clearer you can make that both in terms of like your internal mental model and how easily you can see that by reading the car i think you know the better you become as a shiny developer yeah it's it's definitely a learning journey for sure um lots of things i learned the hard way especially as you start mixing the concepts like reactives and observers and trying to figure out okay where do things launch at what point and there are nice tooling available that you allude to in the book as well on drawing that kind of reactive flow even in your head kind of on paper at first but then verifying are your assumptions actually correct with things like the react log package that uh very slicky has been offering so it's definitely getting easier to find those issues but boy that's the biggest debugging nightmare sometimes it's not so much as an error but it's like it's not firing at the exact time i want it so that's a it can be a frustrating experience but i think we've experienced um it come we experience a reactivity it hopefully gets a bit smoother i think the other mistake that's like one of the most common mistakes i see people doing is putting too much like non-reactive code in the server function like if you've got to do like a lot of complicated you know like data cleaning or data analysis it doesn't actually depend on reactivity you're better off like pulling that out into a separate function so that your serve like your server app almost becomes like just the connections in the reactive graph yes and then you know like the the more you can kind of condense that down and make it like a very crisp declaration of the reactive graph the easier it is to understand what's going on yeah i i wish i could mark like a hundred stars next to that one comment because too many times i see people put these absolutely massive server files just and inside reactives it's it's so hard to decompose like where the problem really resides where if you can factor out that back-end logic and dedicated functions then it's much easier to just debug at those points when you need it but then to rule out if that's a problem or not and i think that's another thing that comes through experience and more like you said kind of the software engineering principles but certainly when i started with shiny it wasn't natural yet but now the value of decomposing and making things more articulate in your intention of where you want to use back-end logic and make your server files whether they're in modules or in the overall server construct more easier to understand and easier to reason with it can only lead to better things yeah yeah and then those skills like you know they transfer to basically any place you're writing a code like the better you can get it kind of extracting out like a you know clear concise idea and like isolating that so you can think about it separately but you know the better a programmer you become yeah right right

your server app almost becomes like just the connections in the reactive graph yes and then you know like the the more you can kind of condense that down and make it like a very crisp declaration of the reactive graph the easier it is to understand what's going on

Creating effective reprexes

and on that thread of kind of like learning and maybe getting help along the way i was again very excited to see that you have a great section in the book about how you can effectively get help with shiny and this is where the concept of reproducible examples come into huge play here now for me personally shiny has always been one of the more challenging areas to develop a reprex for but maybe you could you could um touch on if you don't mind maybe on top of what you're authoring in the book for reprexes what additional advice do you have for shiny developers to help create effective reprexes when they're facing issues with say reactivity or any anything like that yeah just

i think just to emphasize that like to me like the reprex is the tool that i use like if there's a problem you want to like isolate like figure out exactly where the problem is yes and like the more the more you can kind of like throw stuff away to get to like the heart of the problem the easier it is going to be to figure out um like what's gone what's gone wrong and like as the author of the app like you are more suited you can do that more easily than anyone else because you kind of know what's going on whereas if you want someone to help you and then you give them like a thousand lines of code and like one of those lines is a problem like it's really difficult to figure out what's going on and you know it's going to be frustrating when you make your first reprex like it's kind of slow and annoying but it's like it's absolutely a skill you get better at it over time and then just you know it makes it easier for other people to get help you and then you know you discover like 90 of the time when you make a minimal reprex you're like oh now i know what the problem is like it's obvious right like you don't need any like sophisticated

it is something like I enjoy, like here's a hundred lines of code, now I'm going to get it down to like the three lines of code that actually illustrate the problem, and you know that that's kind of like a fun little game. Yeah, the learning journey that I'm on is as I develop these apps with say 5, 10, maybe 20 modules inside, and they're all kind of networking with each other, passing things back and forth, that when something goes wrong in one of them, how I can isolate it to like maybe that one specific module, but find the best way to do a reprex that cuts off the fluff that it doesn't need if I want to post to say RStudio community or somewhere else, and admittedly that's the part I'm probably the weakest in, because when I deal with issues as shiny at my day job, people usually go to me for the shiny questions, so I don't have all first the expertise we're trying to develop, but then if I want to get help, I got to figure out how to decompose a lot of that other unnecessary logic that isn't really impacting the problem, so I'm going to be practicing that, but that's a journey I'm on currently. And I think that also helps you like as you do that more, like you'll get a better sense for like when you start designing something, like there's a little voice in the back of your head, like am I going to be able to like debug this later, and like could I make a design a little simpler.

Right. I forget where this comes from, but there's a kind of a saying like it's always harder to write, it's always harder to debug code than to write code. Interesting. So if you write code that is like you at your smartest, you're not going to be smart enough to debug it. You can't like write code at the peak of your abilities, because you won't be able to debug it, you've always got to like kind of claw it back a little bit. So if something goes wrong, there's a little intellectual headroom for you to deal with it. Yeah, I had a situation with a colleague yesterday, we were debugging a Shiny issue, and I think he was very smart in how we got to that piece, but then trying to decompose why something wasn't firing, yeah we ran into all sorts of hurdles, and I think we walked into that one whether we liked it or not.

So if you write code that is like you at your smartest, you're not going to be smart enough to debug it. You can't like write code at the peak of your abilities, because you won't be able to debug it, you've always got to like kind of claw it back a little bit.

Collaborating on complex Shiny apps

So a couple more things I wanted to touch on is from a developer perspective, what we're seeing now especially in the industries that we that we see Shiny prevalent in, is that we're often working with multiple teammates on even a single Shiny app as they get bigger complexity, may have different parts of the logic or more for like ingesting data versus doing some analytical workflow maybe with HPC technology. And this actually reminds me of the example that you outlined in the module book where I was networking all this different workflow together, and in that case I was the sole developer, but now I've got sometimes I got multiple members helping me. But maybe you could let us know from your experience in you know developing the tidyverse of your team at RStudio, maybe what are some thoughts about enhancing the developer type of collaboration as you're engineering these more complex apps, and maybe you could talk to us about.

Yeah I think there are two things that feel particularly important to me. The one is like trying to like decompose things as much as possible, and then just having a few kind of connection points between them so that people can work like largely independently on different paths at the same time, and you know that they're only connected between these like few ways so it's relatively easy to like iterate on one thing and not break everything else. And then that's I mean that is I don't know that's like maybe the hardest kind of software

engineering skill, like figuring out how to decompose things into independent pieces.

Right. And you know you just you do it wrong for a long time and then it you know slowly you get better at it. I don't know I wish I wish there were some shortcuts but. Yeah I don't know Amoeber.

I know for me when I knew an app was probably going to be bigger than I initially anticipated, I would even use modules even in the case where maybe that module is called once because that's still a separate piece of code that another developer could work on and I could work on something different but then I could be the one that kind of brings it all together and we can yeah do that in parallel. That's certainly been a huge win for me. I think the other part is like really kind of investing in shared conventions so that you can like well like most of the time you stay in kind of you can you know you stay in your lane but if you do need to like touch someone else's code like it's very natural like it feels like you know it's your code uses the same style it uses the same conventions you can kind of drop in and you know basically understand how it works.

Right. And I think that's you know that that's something that we have kind of you know I think in the tidyverse team like that's the things we spend the most time kind of discussing are those like big picture design ideas so that while by and large everyone kind of maintains their own set of packages anyone can kind of like jump in and easily make a fix they know where the tests live they know what the the policies around naming things are they know how to document they know what the basic style is and it just makes it much much easier to be collaborative like you don't want to force force that kind of tight into linking where people have to be constantly kind of communicating back and forth in order to get anything done but when people do want to collaborate it just it's just easy it all feels like one you know it all feels like the code has all been written by like one person.

Yes. Even if you might have forgotten how some of the code works. And that probably was my weaker area in a recent project is we had the whole like decomposing of these different distinct components but then they took a completely different approach architecturally than what I would do in certain parts of the app and it was difficult to kind of reconcile when I brought all that together to understand what their reason is for that kind of logic and their conventions on like variable you know derivations and everything so that's something I'm going to try to do in the future too is have kind of a unified you know developer kind of guideline developer syntax that we all can can follow through because it is more difficult to debug when it's written from a completely different perspective than what you're used to.

Yeah and I find that the other thing I've been trying to do more of lately is when I make like a small breaking change in a package I will do pull requests to all of the packages that it affects. Ah yes. And that's been really interesting I think that that's been really useful at building like my my empathy with developers because like I'm I'm also like as well as breaking and I'm also doing fixing but it's also just it's really useful to see like kind of what are the things that are working well like what are the things that we've done well about like explaining how to develop packages and what are the things that we could could do better at because there's there's some packages that are just really hard for me to fix because I'm like you know I you know I don't know where the light switch is kind of it just makes it like very difficult to like quickly figure out like what the problem is and fix it.

Community and the Shiny ecosystem

Yeah that is that that's excellent advice like I said I'm going to take a lot of those learnings as I develop more Shiny apps in the future. The other thread I wanted to hit on is more the community aspect so as we've seen the tidyverse gain popularity and and gain traction in the R ecosystem you have done an excellent job with your colleagues at RStudio on making the onboarding of getting involved with the tidyverse easier I think of after the previous RStudio conferences having like the tidyverse developer days where we all could get together and and share ideas on fixing issues but the Shiny community I think is also gaining a lot of traction as well so I'm just wondering do you have advice that those maybe getting involved with Shiny on ways they might be able to connect with others in the Shiny community and learning from each other from your from your perspective?

Yeah that's interesting so I think Shiny yeah there's sort of a I think one like one thing that's different with Shiny is like how quickly it feels to me like the ecosystem of packages around Shiny has grown. Absolutely. You know there's like you know dozens from you know more than like dozens 50-60 packages that like more that even that they kind of build on Shiny in some way. Yeah you should see the description file one of my bigger apps that probably has like 45 packages and more than half of them are Shiny related.

Yeah so I think that that somehow that makes it a little like the tidyverse this sort of a much kind of like stronger like locus of control that's going to seem to like you know around my team whereas I think Shiny is much more diffuse so yeah so I think that yeah that's sort of interesting and I don't know what like you know probably ask what we could what RStudio we could be doing to kind of bring that community together and to keep people you know talking to each other you know discoverability pretty quickly becomes a problem like maybe there is a package out there that will really help you solve the problem you're working on but how the heck do you find it and on the kind of flip side like as a package developer like what should you be doing to market your work since you've put all this time and effort into it how do you help people actually discover it so it can help them.

Yeah it's it's a it's a big issue but I'm certainly doing my part with things like this developer series to highlight some of these but certainly I'm only one person so hopefully others will will be along that journey but I know I've benefited from the networking I've gotten at various conferences that I've maybe they're industry specific but also just putting things out there of like example apps and certainly what RStudio has done with the Shiny contest has been a huge benefit as well so you got to see how people are blending these different other packages together then you can kind of see unified themes on ones that are appearing maybe more common than others and you can kind of grok them what are the ones that you think you should be investing in but it is I won't say a wild west but there can be a bit of variation in what these packages are doing so you definitely need to have a more honed in on perspective and make sure it's useful for the project you have but from what I've from my experiences the community has been very welcoming they love when we report issues they love when we can contribute pull requests to enhance a package and I've certainly done that for a few of them so I hope it just grows over time but I think it's an exciting journey to be on for sure.

Yeah and I found that like you know there are a few there are a few packages I talk about in the book and you know basically whenever I start using a package I discover like a lot of little things that kind of bug me and some of them about 10 issues or something and sure developers are like really happy to kind of dig in and fix them which is which is awesome.

Yeah and I know that in in now I'm in a particular industry where I have to keep a lot of things close to the vest so to speak but as we start to build examples that mimic real world usage that's when we get the opportunity to do more networking with other developers and share our ideas and maybe for future enhancements and certainly yeah I think we all can can benefit from that.

Mastering Shiny: timeline and open publishing

So I guess to kind of wrap things up here the book is still in progress but what's your expectations for when it will be finished and in the hands of in paper form so the goal is to like finish it um next by the end well the goal was to finish it today but I did not make that goal oh today really.

Next Friday so that basically means like I'm kind of done with the all the big picture writing okay then it goes over to O'Reilly for like the you know they start assembling the book and making sure all the images the right size and do a bunch of proofreading and stuff so I think it's like probably under six months away now hopefully I think probably three to six months away I would imagine before you can like get a physical book in your hands um and then you know the content on the website is basically done so yeah well if you've been waiting to start reading it until it's finished wait wait no longer.

Well I'm one of those people I'm an early reader so to speak so it's great to see the evolution of the different sections in place and yeah it's been I just I mean as a kind of an aside the open publishing model that started with things like R4DS has been so helpful to the R community so I'm so thankful that you followed that that same principle here so we could kind of see the development in action and comment on ideas as you were reaching out to the community on authoring certain sections I know you would reach out on Twitter sometimes and say hey I'm offering this section any advice so that's really been benefiting everybody I think.

Yeah I mean I I love it too because it'd be otherwise if you just sit down to write a book and you get kind of no feedback it's hard it's so hard to keep your motivation high and like even when people are reporting problems or complaining about things it's still like a positive to me because at least like they read it and they cared enough about the plan and that like keeps me keeps me motivated so yeah yeah getting that feedback so helpful.

when people are reporting problems or complaining about things it's still like a positive to me because at least like they read it and they cared enough about the plan and that like keeps me keeps me motivated.

Leveling up your Shiny skills

So I guess one last kind of question to the Shiny audience is do you have some last bit of advice on top of course reading mastering Shiny of how users can kind of level up their skills as they start to transition from like that basic familiarity to developing more complex you know applications.

I think the best advice I can give is to make sure that you are setting aside time to kind of improve like the quality of your app like not so that the what it might do stays exactly the same but I think the quality of your code increases and like making sure that like like you understand why that's important because it you know increases it allows you to keep the pace of development up over time and then also selling it to like you know your boss or whoever like you know I'm going to be working on this there's not going to be any like visible outputs but you know the ease at which I can maintain this is going to improve and my software engineering skills are going to improve and if you can like regularly invest in your programming skills and your software engineering skills that's just you know that's something that compounds over time you get better and better and better.

if you can like regularly invest in your programming skills and your software engineering skills that's just you know that's something that compounds over time you get better and better and better.

Yeah well that's that's an excellent way to wrap that piece of it up so in closing if anybody wants to get in touch with me either about the book for about anything else what's what's the best way they can get a hold of you twitter twitter and github nope that's that's everywhere I've been interacting with you a lot yep that's great well um Hadley again thank you so much for joining me for this episode again on behalf of the overall Shiny community we are so excited to see your efforts and you know helping us on the journey of learning Shiny with this excellent resource and I'm sure we'll be hearing more about you in this space even in the future so yeah thanks Eric all right great to talk of you Hadley we'll be back right after this yep.

Closing remarks

Well that was extremely exciting to hear what Hadley has in store for us with with this excellent resource and the advice that he gave us from his wealth of experience with obviously authoring the tidyverse related packages and coming at this from the practitioner standpoint and this software development mindset it's just such a great time to be learning about these concepts in Shiny because as I alluded to in the talk a lot of these resources simply were not around when I first started with Shiny so you're you're coming at this at a great time if you're if you're kind of new to this journey.

So if you want to hear more and be up to date with the great community of Shiny certainly the best way to get in touch is to subscribe to this very YouTube channel use the little links below and all the resources that are mentioned in each of our episodes is posted on our official website of shinydevseries.com and also keep an eye on our RStudio community we'll have some posts there in the near future as these episodes are being published and I have lots of plans for future guests this year and you'll be hearing about that across all the spectrums of Shiny development and it'll be really exciting and of course I'm going to be continuing my live streaming efforts so I kind of learn out loud with all of you and try some fun projects and certainly I might do some more on that little fun racing initiative I've been a part of so okay so once again feel free to get in touch with me I'm on Twitter at at thercast and also like I said head to shinydevseries.com you'll have a contact page you can file your your request to as well and I'll be glad to talk with all of you all right so that's going to wrap up this episode the Shiny Developer Series and we will see you next time bye-bye everybody.