Positron workflows that make life easier | Andrew Heiss | Data Science Lab

Transcript#

This transcript was generated automatically and may contain errors.

I'm excited to introduce our featured guest today on the lab, Andrew Heiss. You probably know him from his blog. I certainly do. Andrew, would you like to say hello, introduce yourself?

Hi, this is exciting seeing all these people. So I'm Andrew Heiss. I'm an assistant professor at Georgia State University in Atlanta. I do our stuff. I teach data visualization, causal inference type stuff. If you're on social media, you've seen me there. I was on Twitter for years and then it died. And so now I'm on Blue Sky and kind of LinkedIn, but mostly Blue Sky and have a blog. Lots of people use it. I have lots of teaching websites that you may have seen. Lots of you have copies of them and that's awesome. I always stumble on random class websites and it's my stuff and it's always fun to see that. So copy my stuff, everybody. This is cool.

You heard it here first, copy his stuff. I feel like that's the open source community, we're all about it. All right, Andrew, what are you going to be showing us today? And if you would like to go ahead and start sharing your screen, you can do that at any time.

Yeah, let me share my screen. So in general, this is my first time ever actually being at a data science lab. I've been to a couple of Hangouts and those are more formal, but this is just showing cool stuff because lots of people ask me all the time how I do stuff so fast. And one, I don't, it just looks like it. But two, I have lots of cool little programs and extensions and things that I use a lot. And so I switched to Positron in September 2024, a few months after it first came out. And so I've been using that full time for the past almost 18 months now. And so I have lots of cool little tips and tricks that I've stumbled across over the past year and a half of doing that. And then just other general neat things that make life easier. So that's kind of what I'm gonna show you all today and answer whatever questions you have. So excited.

So let me share my screen here and switch away from Positron. Oh, so I actually, in the spirit of just sharing everything, if you go to my blog, now there is a post with all of the links to the things that I'm going to show. And so I also put that in Discord just now. So you can get to it directly that way too. But also just ask whatever questions you have. So in general, like, Libby and I talked about this before, kind of the general outline for this. And so what we wanted to start off with was just general Positron stuff, how to find good extensions, how to do some fancy neat things with like the connections pane in Positron, other things like that. And then we'll look at some other cool workflow things that work both on Mac OS and on Windows. So if I'm a Mac person, but if you're a Windows person, you can still do all of this cool stuff too.

Evaluating and choosing Positron extensions

So first I wanted to talk about the extensions. Because this is something I've noticed a lot. I still have not switched to Positron when teaching R. I'm still using RStudio in part just because VS Code slash Positron is kind of a bigger beast because it works across like all languages, not just R. And so with students, I find that it's easier to just stick with RStudio. And then near the end of the semester, I point them to Positron and say, hey, this is a thing. And so with RStudio, that ecosystem is like related to the general R ecosystem. So all of like the RStudio add-ons, we tend to like inherently trust them because it's people in like the R community making little add-ons like ggthemeassist and other neat things that let you do cool stuff. And so I've rarely run into people like worried about installing extensions or add-ons in RStudio. But I have run into that with VS Code slash Positron, in part because that's a huge community. It's people who do like Rust full time or JavaScript full time, and then they make their own extensions. And so I don't know any of those people. And there's no like CRAN that like vets everything.

And so I'm not necessarily worried about like malicious code showing up in the different extensions, just like how well the extensions work. And so one thing I wanted to show really quick is just kind of how I can tell if stuff like if an extension is good and safe to install. I have like a few heuristics here. So like this right here, this better comments thing, it's on so the OpenVSX extension store is like the open source version of the VS Code extension store. But all of those extensions live at GitHub. And so I'll check like if I come across something like this better comments thing. I look and see what it does. It's got cool screenshots and stuff. That's neat. And so they'll do special syntax highlighting if you have like a to-do comment or other types of comments. But then I also check to see if it's highly rated. The OpenVSX ratings are like useless, because very few people actually use this thing. So it's got like, I've never used it. Yeah, like the five people have clicked on five stars. Neat.

But like tons of people have downloaded this thing. So sometimes they'll even look at the repository or the the extension on VS Code's store just to see what people there are saying about it. Because the OpenVSX world is kind of dead. Because nobody comments there. But also look at like the GitHub repository for it just to see this hasn't been updated in four years. It's got a bunch of issues still that are fairly recent. And nothing's been changing. But all it really does is change syntax highlighting for stuff. So I'm not super concerned if it's not super updated, because it doesn't do much. If it's something more important, more complex, and it hasn't been updated for a while, then I'll just like not. I'll still like install it and play around with it. And generally, it'll be like really buggy if it hasn't been updated for a while.

But those are my general heuristics. I'll look at like the OpenVSX page, check the GitHub page, check the VS Code version of it and see what people are saying about it. And then if it looks cool, I'll install it and play around with it.

Recommended extensions: air, better comments, and pasThem

So some of the good ones that I've found, I put air here just because it's neat, but it's built into Positron. So this is the automatic code formatter and linter. Some people actually have there's a setting in Positron where you can have it run every time you save a file. I don't want to relinquish that much control to computers like reformatting my code every time I save but some people do and that's cool. What I mostly use it for is like if I have code, let me just grab some code like ggplot and then AES. So I'll do something like AES x equals something plus geom point. Okay, so that's like horrific R code there. If I just select it and then do Command Shift A, it will reformat it using air and follow all of the fancy tidyverse rules. There are other settings you can change with air. The documentation shows how to do it. I haven't messed with any of those. I'm happy with that and it works. It has cool heuristics too. Like if it sees that I have like the data set on a new line and then I run it, then it assumes I want all of the arguments to be on new lines and so it'll update that automatically and so air is really cool for that.

Better comments I already showed like if you have some comments here like neat this is a comment and then later I say to do this is important. Notice how it's like orange and that just automatically got picked up because I have that better comments extension turned on and so it recognizes a bunch of other things. I basically only really use to do because I forget what the others are but I should probably eventually learn what they are because they're cool. I used to do a lot so this is really useful. I want my to do's to stand out or I want them to be put into a list for me somewhere. Yeah and there are other extensions that will give like a little pane in the side here that will list like all your to do's. I think one is called to do tree. I have that on for a while but what I ran into there is it picks up everything in a folder and so if you have like quarto extensions or renv actually this ran in this was an issue because you have like whole copies of like our packages and so if other people are using to do's in like extension files then those also appear in like the list of everything you have to do.

Cool. So some of the good ones that I've found, one called pasThem or paste them I don't know what it's officially called. Um this is so cool and very few people know about this so get ready for some magic. So if you like select some like a spreadsheet from excel or just like copy a table from html or something and then if I come to like Positron here let me move this out of the way and then if I paste it it pastes it like this it's no longer like uh tab separated or anything. But what I can do is if I open like the command palette thing and do pasThem I can here show all of these pasThem so I can have it paste as a julia data frame or a python data frame or a markdown table or an r data frame and if I choose our data frame I can say I want it as a table or a triple or a data dot table or polars or base. So let's do a triple there it is it's automatically a triple.

Um this is so cool and very few people know about this so get ready for some magic.

And automatic so yes then let's do that thank you uh miles mcbain for being the data pasta pioneer here and then I think that pasThem was done to like mimic data pasta by anatoly I think is his name so thank you anatoly for giving us that because if you haven't used data pasta it is an add-in for RStudio specifically for RStudio. Add-ins are like RStudio's version of these extensions and Positron. And it's magical and I love it I use it all the time and so when Positron first came out that was one of the things I really missed so pasThem fantastic.

Notice how this window is blue and if I open up a different data viz thing so here's my fall one it is red that's so cool so the the reason that works is because there's an extension called uh peacock and that lets you set project specific colors to your windows.

Connections pane and DuckDB

So over on the side here you have this connections panel um RStudio has it too um and I find this is really helpful if you ever have to work with databases which I rarely do. Um so all of my stuff is just like academic modeling with like csv files and excel files um but occasionally I will have to do stuff with databases like connecting to an sqlite database or a mysql database or a spark thing or whatever. Um and the neat thing about Positron is it lets you like navigate the database directly from from Positron itself so you can like see what the tables are and see what's in there. Um which is super useful um because I like some people love the command line for like navigating sqlite databases I don't know how to do that um and so I like like clicking on stuff and sorting stuff and seeing what's actually in there and so the the connections panel lets you do that.

Um so on my little website thing I actually have a blog post tutorial showing in general how to do it. Um I have these these two real life use cases where I've done this before because again I don't typically do database stuff. Um I did help the state of idaho make an election reporting website back in 2024 um and for that they have like their own proprietary database system that I couldn't access and so what I'd ask them for was like the column names in the database so I made my own like fake version of it locally that I could like play with and it was the same structure and that helped. Um but where it's really helpful is I have this current project that has a csv file that is zipped and when it's zipped it's 10 gigabytes um and there's like no possible way to load that into r um or you can but it like blows stuff up and it's bad. Um and so one neat thing you can do is you can shove the csv file into a duck db file which does some magic compression stuff um and makes it work really well so once I get that 10 gigabyte file in duck db it's only like 600 megabytes ish. Um and then you can do like queries with it which again I rarely use sql queries but there's a db plier that lets you use dplyr stuff to get stuff directly out of it um which is cool.

Um so let me just show you what that looks like in practice um what the connections panel looks like and so I have the code here on my little website thing I'm just going to copy it and walk through all the different pieces you can see what it looks like so we'll just make a new r file here so tidyverse great so what I'm going to do is connect to a brand new duck db file if you've never used duck db that's okay um because it's just it's like my s it's like sqlite that's like a single file thing that holds stuff in it so now I have this cool file here called my fancy database dot duck db. Um it's just some sort of um file I don't even know what it looks like I don't know if it's binary or plain text it's something um doesn't matter because what I can do now is I can copy to it and so I'm going to stick gapminder in there because why not. I could stick diamonds in there I could stick whatever I want in there um and over here in the connections panel I can now click at that and then look inside duck db and kind of twirl down here and now I have gapminder inside that database and so I can click on this and it shows in the data explorer so I can actually see all of the data inside there which is cool. Um you can use db plier to get stuff out of it so I can use this this table function that will connect to that database pull out that gapminder data set um this will actually run filter as an sql command like it does some behind the scenes translation into sql and then collect will bring it into r as a data frame so that is now gapminder 2007 but the way it worked was grabbing it from the database through some query collecting the data and bringing it back in.

Raycast: a supercharged spotlight

So next is um general non-r workflow stuff that makes life easier. Um and so I I did mention already like raycast like if I mash all these things down and press p this brings up all of my projects and so I can open that so what raycast is is if you are on a mac you have if you press command space on your mac it opens up spotlight which is like the built in way of like searching for things on your computer or um opening folders you it's a calculator too if you type one plus one it'll show you two. Um and so you can do that with spotlight um raycast is basically spotlight but fancier like super fancy. Um it's free um if you go to raycast.com um you can download it they just released like a beta version for windows like a month ago um so I'm assuming it's cool I have not tried it yet because I don't use windows so I can't even try it but it's probably cool. Um so raycast lets you do all sorts of neat things they have a huge presence on youtube um the way they make their money it's free but they have like a pro version where you can like sign up for different ai models and does other ai type stuff um and so that's how they make their money but I don't pay them anything because I don't use any of their ai stuff.

Okay cool so everything you're going to show is on the free version yeah everything here is free and so yeah if you check out this 101 things you can do with raycast it it takes like five minutes and just goes through really quick so like here's a calculator one plus one neat um you can say like next tuesday is january 20th neat so like that but you can also search for stuff like if I want to open up minecraft there it is um that's cool so it's like a general thing for searching your computer.

Yeah is this is the like command option control p is that something you decided or is that a recast that's something I decided so that's that's where raycast gets magical is like on its surface it looks just like a regular like application searcher or file searcher like spotlight but if you go into the settings for raycast it has all sorts of like extensions that do other things and you can set global global keyboard shortcuts for them. So like I have this setup where if I press command option k that will paste whatever is in my clipboard as plain text and so if I'm copying from like word and there's bold and italics in it that will strip it all and just put it as plain text wherever I am. Um and so that's that's where I'm setting these hotkeys here so this little diamond thing that's uh that's a short version of control shift option command like mashing all of that stuff so if I do mash all that stuff and then press g it'll bring up this little thing where it will search my github repositories so I can search for tidyverse and then that will open up a new tab.

Now Adrian and I just had the same reaction which was like very large eyes and slightly upset because that is so cool all it does is like there's a in github if you type user colon your name and then whatever that will search your own things but I always forget that and so yeah I just like mash this press g and then if I remember like I remember I had gg halves once and so I'll do this and say oh there's how I use gg halves and so I can like find it there.

Um that link to searching on github like I have the magic incantation to do that a bunch of other stuff so there's the project manager smashing everything pressing p there's a project manager extension you can install and it works. Um there's some cool scripts so the other cool thing about raycast is it can run any arbitrary script um and so like I often like when I'm setting seeds for like analysis I like to go to random.org and grab a seed from there just because it feels more random and uses atmospheric noise or whatever so if I come here and type seed and press enter it just grabbed it and copied it onto the clipboard so now I have that as my seed 8 9 3 0 0 7 so that just grabbed it because it ran some script to do it. So I actually have a link to all of my scripts here if you go there you can see all of the cool stuff that's there so there's get seed um I have a thing that toggles my little ring light here that's on my webcam so I can turn it on and off.

Um so if you notice this thing up here in my menu bar that shows that I'm currently in this meeting and it knows that because raycast can put like your upcoming events up in the menu bar and it's cool. Um and it's smart enough where like if it's right before the event and you press like command space to get to it it will show that current event like the upcoming event there and then I can just hit enter and it'll open zoom or webex or whatever. Um it has a built-in clipboard history thing which is magic so you can go back and see everything you've copied and go back and grab it so you don't need like a third-party clipboard history manager it's in there. Um it's got an emoji picker so I can like do this type of thing and search for emojis or whatever which is cool. It has this is cool too if you search for a color picker it has a built-in color picker so I can say what that color and it copies.

Espanso: system-wide text expansion

Okay so the last little thing here is weird um this cool program called espanso which is a separate program from raycast that does a completely different thing. It is also free and um maybe open source it's cross platform it works on windows and linux and stuff. Um if you've ever heard of text expander I used to use that for a long time it's the same thing but free and open source. Um and so um basically you type something like in that little demonstration they typed colon date and it automatically expanded to the current date you can do that with all sorts of cool stuff. Um and it's it works across the whole system so like Positron and RStudio have ways of like setting snippets of code so if you type something it'll like put code there that's helpful I don't like doing those because I'll often be in different editors and I don't want to like set up one snippet in one place and a snippet in another place.

Um and so I I don't do that um so what I do instead is if you look at this link here these are all of my espanso shortcuts um and there's all sorts of like really cool things um so for example let me just show a few highlights um there's like a billion and this again is just like the accumulated cruft of working with text expanders since 2009. Um and then I didn't then they moved to like a subscription thing and then I didn't want to pay for that and then espanso existed so and they're free. Um raycast has its own system I just it I don't like it as much it's not as fully featured as espanso so I don't use it but you could do that too. Um so for instance if I'm like typing stuff and type semicolon d t it expands to the current date if I do semicolon d a t e it does that date if I do semicolon d z it does a date with the little t thing inside for like a blog post like um timestamp. Um so you can do all sorts of like cool stuff that's all just keyboard shortcuts that I've set.

I am a typography nerd and so if I'm writing something like one minus one that's the wrong minus sign um and so I have a thing where if I do semicolon minus that's the correct minus sign which uses the same the crossbars at the same level as the plus sign so they line up. I love it that has always bothered me but I if you do like can't be bothered to like learn the alt commands yeah the windows there are alt commands on a mac there's lots of alt commands like you can do option minus and but not here um option minus outside of Positron will do like an n dash. Um so what I do here is if I do semicolon e n n that adds an n dash so it's kind of a workaround for working in Positron there. Um if you have like an x here I have it set to do the actual multiply sign if I do times so that's the true multiplication sign. So stuff like that or if I'm just like typing I study institutions which is a really hard word to spell if you're writing fast actually study authoritarian institutions and that's like a lot of words and I will write authoritarian institutions and it's like awful so I have it set where if I type a u t h n it expands to authoritarian and then I s t t n institution. Or like n I have n triple because I can never spell that word so it expands to entrepreneurial.

Um so all sorts of cool stuff it has cool menus too so if I start a brand new document and want like yaml metadata for like a Quarto file I can do semicolon semicolon qmd and it popped up on the other screen it brings up this cool menu thing where I have different headers that I can potentially use like a basic yaml header there it is and so now I can insert stuff. Or let me do this so if I do qmd again let me show it over here I can say full academic Quarto yaml header it got confused because it's on a different screen it's okay the joys of screen sharing. Oh it's messed up but it would put a bunch of stuff there. Um yeah so all sorts of cool stuff if you look at the website there's other things oh this is another cool thing is if I type call out it brings up this cool little thing that I've made so I can add a Quarto call out like a warning that is collapsed that is the simple appearance without an icon so I can just type call out that is the simple appearance without an icon submit that's all there and now I can add my little call out warning stuff. So I have lots of little things for adding uh Quarto specific things like if if you're making like a reveal js slideshow and you want multiple columns I can do col and there's my column divs.

Um so all sorts of cool stuff it has cool menus too so if I start a brand new document and want like yaml metadata for like a Quarto file I can do semicolon semicolon qmd and it popped up on the other screen it brings up this cool menu thing where I have different headers that I can potentially use.

This is so cool this is the stuff that I never remember how to do and I almost always have to go copy paste from another one of my documents. Well we have some questions that have rolled in through the discord we have three minutes left let's see if we can hit some answers for things really quick rapid fire.

Is raycast a replacement for spotlight and bartender it's a replacement for spotlight it's not quite a replacement for bartender so bartender on a mac is um if you have a menu bar with lots of things in it bartender lets you like collapse them into like a hidden menu bar bartender is expensive okay and but there is a thing called ice menu bar that is like an open source menu bar thing for mac os so this is an open source bartender and that's what I use now it's great.

Cool someone also asked logan asked what browser are you using I think you're using arc using arc so I will say oh ryan put it in arc.net as well. Um Sharon asked is there an easy way to see all of the shortcuts that you have already set up I think there is um so all the espanso shortcuts kind of um so if you look at this all my shortcuts and settings that's a git repository so espanso is cool because it's all plain text configured. So if I come to like match here so like here is um programming stuff so if I type like semicolon wm it will make this warning false message false Quarto chunk metadata thing. Um there's nothing magic about the semicolon thing some people do colon some people uh won't have like a prefix I just like the semicolon thing because I don't typically write like semicolon fig w like that's not a common thing I type um so it's not going to accidentally trigger it so everything is in this git repository there. For raycast if you filter your stuff to show only customized then you can see all of the stuff that you've set and see all the keyboard shortcuts there.

Amazing Jared your question I don't know what it means but I'm gonna ask what shortcut have you set for an intera bang oh I don't have one um but I should so that's the the combination question mark exclamation mark like in one character. Um so oh okay I get it intera bang um how David under asked how is he triggering those word replacements so that is because espanso is a it's just running up here in my menu bar on windows it would run down in like the taskbar thing and so it just watches what you type and so if you trigger any of these things like this r stop or semicolon r stop will add this knitter knit exit thing. Um and so that will trigger it if I have I have this long words thing so I've got lots like multinomial can never spell that so I have it semicolon mnl um authoritarian I don't have a semicolon there I just like start typing au th and and then it magically turns into authoritarian characteristics that's a tricky one so I type cars and it magically changes.

Um so yeah you can see all my stuff so these are just like code snippets but they're for words all right well we've we've hit the top of the hour I know people are gonna have to jump this was amazing Andrew thank you so much for coming this has been recorded it's going to be on youtube eventually thank you thank you thank you I hope you had a good time everybody stick around in the discord server continue to ask questions answer each other's questions um and we'll see you all next week on tuesday. Andrew would you like to say goodbye thank you so much yeah thanks for coming everybody this was great I can talk about this stuff like all day this is it's my way of procrastinating is I could listen to it all day. Okay well go read Andrew's blog he so kindly packaged that all up for us and we'll see you this thursday on the data science hangout we have Mina uh Çetinkaya-Rundel if you know Mina you know that she's amazing come see her on thursday and then we'll see you next week at the lab bye everybody bye.