Lightning Talks (Tuesday) with Andrew Mshar
Published October 26, 2026
This video features Andrew Mshar at DjangoCon US 2026 in Chicago, Illinois, USA.
This talk was presented by Andrew Mshar at DjangoCon US 2026 in Chicago, Illinois, USA
https://2026.djangocon.us/talks/lightning-talks-wednesday/
LINKS:
Follow Andrew Mshar 👇
On GitHub: https://github.com/programmylife
On Mastodon: https://fosstodon.org/@programmylife
On X: https://x.com/programmylife
Website: https://programmingmylife.com/
Automatically transcribed, so expect mistakes in names and technical terms.
Speaker 1: Also, for anyone giving a talk, there is a five-minute timer up here that you can use that you'll be able to see. The slots are all filled, so we'll get started here in just a minute. I'm just gonna go through the names to see who's here Uh Charlie's up first, and then we have Francisco, great. Adam, fantastic. Ayush, I saw you, yep. Annette. Okay. Uh AJ. Okay. Uh Mariana? Great. Uh BarTek. Sure, Bar Tech will be in and Josh. Great. Okay, so we got most people. All right. Charlie, go ahead.
Speaker 2: All right. Alrighty then. So my Lightning Talk will be relatively short, even by other Lightning Talk standards. So , is it this plane? Okay, it's plain. So, for the for those who don't know me don't know me quite well, I uh I contribute quite profusely to a wide variety of Python project of Python projects, including including but not limited to the Python interpreter itself, low-level scientific libraries such as such such as NumPy and cross-language bindings such as PyO3, which allows Rusk which allows Uh Python c Python to call ex to call external Rust libraries and vi and vice versa. And as you can tell Sometimes sometimes one sometimes one is tempted to immediately open immediately open APR on e on an incremental improvement to to an existing problem.
Speaker 2: Without without uh without ha having re having reached uh how do I put this a a broad consensus with with with c with core project maintainers on on this on your s on what your solution being the best the the best one forward. So this is this is mine. So my situation was that was that that w within a Python inter within the Python interpreter, uh there there there are build there are build flags available that pass pass through its uh configure scripts to uh uh to uh to build the interpreter with address with address and address sanitizer such as uh thread sanitizer, address sanitizer and And m and memory s and memory sanitizer. So my PR was to add even more sanitizers supported by
Speaker 2: is more sanitizers for for the internet to be built with. Uh s such that such uh actu actually let me wind back. Uh both Clang and G C C support s support uh more sanit s the sanitizers than that than the than the existing three available for the C Pont interpreter. There was my AR and as you can tell I got scop uh it was not only rejected, but got generously scolded by Guido van Rosum. On a show of hands, does anyone know who Hido van Rosum is Yep, just as I expected. So this PR in question if I can this open it Uh oh sh I was for I forgot to
Speaker 2: uh if I can end exit from the slides, just Okay, end slash show. Okay, now you should be able to see my screen. Okay, let's duplicate it. Okay. This is this is the PR in qu this is the uh ish issue in question Well, what flags to be added to the to to the in to Python's uh configure script. And scrolling down, there's this nice comment. Oh, sorry. This one. And yep, that 's
Speaker 2: And yep, that's b that's my talk. There there's my own.
Speaker 3: So hi everyone, I'm Francisco. This is my first time at a DjangoCon. It's also my first time speaking at a conference. So you know because of that I thought it would make sense to talk about the first time I actually read the Django source code and why I had to do that. What I learned in the process and why this specific episode resulted in a new line of documentation added to Django Project. com So just some context. I have recently graduated from college. That was two weeks ago, in fact. And thanks. And
Speaker 3: while I was I was in college, I got this internship at a company called OnSign where I work to this date. And one day I was tasked with updating our code base from Django 3. 2 to Django 4. 2 And you know, I did what had to be done, went through all the release notes, read everything, removed any deprecated features that we were still using. Ran some tests, made sure that everything was okay. Everything seemed okay, so you know, start typing git commit update Django to 4. 2, hit enter, and wham. And what I had never seen that before. So you know, went back, read what it
Speaker 3: said And um it says system check error. And for four of our different models, it said check class method is currently overridden by blah blah blah blah And then I was like, okay, but why can it be overridden? What what is this? What is system check error? So I I did some research and I, you know, found out about this wonderful thing called the system check framework Which I didn't even know existed at the time. And then I thought, okay, well, fair enough. I get it why we shouldn't override this method, but You know, quite a few of our models had have had a check field for a number of years. Why am I why are we only getting this now So I went back, read the release notes again
Speaker 3: with some extra care, but didn't find anything. So I, you know, wrote wrote up my sleeves and started looking into Django's Model Feuds module And you know, I I had never done that before, like looked at the actual code and um it was quite you know interesting and quite cool to get to know a bit more of the internals of this software that I had been using for a while, but Up until this point had only read you know the documentation, not the code itself. And then I found this commit introduced in Django 4. 0, which it it's a bug fix that hadn't been um uh didn't make any of the release notes and it's a fix that made it so model fields uh would now always override any inherited properties or methods with the same name.
Speaker 3: And you know this is what triggered um that system check. And to fix this in our code base, I thought, you know, we can rename the field and update every call site, but that's just too much work. So I just silenced that check. And no and I registered a custom check all models method that basically uh you know makes sure that we call the check class method for our models that would override it. And you know, that fixed it. It worked. It works to this day. And um then I I took a step back and I thought, you know, this should be documented. And I looked up Um Django Docs, field name restrictions, and it says three things.
Speaker 3: A field name cannot be a Python reserved word. You know, makes sense. A field name cannot contain more than one underscore in a row for the lookup syntax, and a field name cannot end with an underscore for similar reasons. And as we can see, the word check does not fit into any of those criteria. So it it should be valid by definition. And then I opened my first ticket and uh basically saying like shouldn't this be documented that you know a field name shouldn't be checked um to which Sarah Boyce replied shout out Sarah I don't know if she's here uh yeah it should be documented And it was. Not by me, someone else was quicker, unfortunately, it would have been a cool first contribution
Speaker 3: Um but now as you can see there's a fourth restriction saying that a field name cannot be checked as this would override the checks frameworks model check method. Uh and that's there because of me. So, you know, what I took from all of this, and I think it may sound a bit obvious to the more experienced people here, but I think it's, you know, good advice for the ones that are starting out like I was when all of this happened is you know you should always read the docs especially Django's which are you know so well written But sometimes the best way to you know truly understand how it works is by reading the code itself. Code by definition is never ambiguous and documentation can sometimes be
Speaker 3: And lastly, before opening a ticket or raising an issue because you know you think you found a bug or some oops or you know that something is wrong. Read the code, understand the code and what is going on, and you will learn a lot in the process and probably make the life of the person that is going to read that ticket, you know, easier because you'll have more information on the issue itself. And that's it. Thank you.
Speaker 4: Hello. I'm Adam Zapletal. I work for Crossway. We are a not-for-profit Bible and Christian book publisher. And we are hiring right now. So if you're aligned with Crossway's mission and you want to come work on Django stuff with me, uh come talk to me or message me on Slack or look at crossway. org for the job posting. Um uh the title of my lightning talk is Friday Afternoon Time. So this is something we do on our team and our department. So part of our weekly rhythm is Uh to have a midday department meeting where it's kind of like a little bit of an extended stand-up where we give an update on the week and talk about the things we accomplished.
Speaker 4: That's probably not too out of the ordinary, but It's it's in the middle of the day on Friday, and the reason for that is because uh we keep Friday afternoons for a very specific purpose, and that is A way to bless our developers, build culture, and kind of ease into the weekend. Friday afternoons are probably not the most productive time for everyone, so I think it's a good fit Slightly inspired by Google's 20% time, but but not not exactly the same. So some of the activities we do during this time is like Sometimes in the year we'll have a developer book club or you can read a technical book. We'd share articles and discuss them. We pick a programming problem
Speaker 4: and work it out and submit it to an online judge and then you know discuss the solutions. We try out new tools or software, whether it's developer tools or a package for our um one of our projects or something, our few new NeoVim plugins. Shout out to all the NeoVim users out there. Okay. Uh nice. Uh and finally, another option is to do open source contributions. And I'll focus in on that soon here. But um I just want to encourage everyone, if you're a developer, maybe consider talking to your manager about having this kind of time, or if you're a manager consider trying this with your developers.
Speaker 4: We've just seen a lot of fruit from it. We're remote teams, so it's a good culture builder. It's a lot of fun. Um and you know, it's produce more fruit than it's cost, I think. So that's one main takeaway The second thing I want to focus in on is the idea of doing open source contributions during this time on work time. Now that's been a a really good blessing for us, and there's been so many uh talks and ideas at this conference about giving back to Django and contributing to open source and that's all been great. And I don't think I heard too much about doing it at work. So I want to pitch that idea to you as well. And you know, we heard imposter syndrome mention
Speaker 4: mentioned and all these other kind of ideas about how difficult it can be and it's true. It's hard to put yourself out there. It's hard to uh take that first step and get into the rhythm of contributing. It's even hard for me as someone who's contributed fairly regularly for a number of years sometimes. But I think there's a special thing that happens psychologically or just in the way we think of ourselves, at least for me, when we're given work time for that um because it's hopefully a little harder to feel like an imposter when you're at work being paid to do the thing that you're doing, right? So one way you can help people to not feel an imposter syndrome is to give them a little bit of work time to contribute.
Speaker 4: And it works, it blesses people all around. Your team culture, the projects they work on. Yeah, it's great. So You bring people out of isolation by doing that, I think. And checking my time. Ominous presence. Um sorry Uh you don't you know all of the ideas in the last talk we heard were great, the last full talk about open source contributing. You can do this as a team as well. You don't have to do it in isolation. So imagine an idea. Let's say you take two or three Friday afternoons. And two or three developers on your team. Maybe the first Friday afternoon you pick uh you take some time and try to narrow down tickets you want to work on the next week.
Speaker 4: You start researching and prototyping the third week, you work together to submit it submit a PR. Just imagine what that can do for your developer culture and then and the projects that you depend on. I think that's about it. Consider trying these things. I think it'll be a huge blessing to your developers. And oh, and It can also be viable to just go review a Django PR like Sarah Boyce was talking about. As a team, you don't have to do that alone, right? That's it. Thanks.
Speaker 5: All right, everyone. So I come from a theoretical physics background, so I got a few friends here that were interested in understanding how black holes work. And I was thinking of a couple different approaches on how to work on that. Uh the first approach was to go through the paper, which I understand can be very boring. So I'm going to dive into well what happens uh from an experience point of view as you enter the event horizon of a black hole. But do not worry, the the closest one is about 1500 light years away. It has some microsecond or nanosecond. Um effect on on our time on earth, which is really cool to think about. But you know, it's not something that we need to worry in our lifetime
Speaker 5: or even after that. Um so really the entire interesting story comes in when you are when you have two Let's just say Alice and Bob. This is coming from a cybersecurity perspective. But uh let's just say that uh Alice goes inside the black hole whereas Bob is at a safe distance. Now uh this now this might be something that you have heard that from a Bob's perspective, Alice would seem to uh get stuck or freezing in time. But in reality, as time passes for Bob, Alice would not only be frozen in time, she would eventually become redder and redder and fainter and fainter at the same time.
Speaker 5: but not completely disappear all at once. Um now what happens when uh you think from an Alice Alice 's perspective? Part-time flow is normal as usual. She doesn't notice a thing because this is not the most dangerous part. So the event horizon is not a wall. There's no welcome sign, right? Like, hey, you're in for an adventure of your lifetime, right? And for a supermassive black hole the crossing the horizon could feel locally uneventful because um it the larger it is the less violent it is so uh if you ever do happen to get close by pray that's a very large one.
Speaker 5: So, but what happens once you're um once you're past the event horizon, right? Then Theoretically it towards the center becomes your future because time and everything starts flowing towards the black hole. Now think about it. This is a very interesting phenomena to think about because I've heard questions such as, well, what if I get enough force to exit out? That's not possible because Your timeline actually starts with you flowing towards the black hole. So even if you happen to get in a force, your timeline is still pointing towards the center of the black hole Right? Which is very cool to think about, at least from a physicist's perspective.
Speaker 5: But it still does not mean I w I would want to be near one, just in case. Um so um there's there's a couple of uh theories on on what a singularity, what would replace singularity, but but to understand singularity I've heard. from most people that they think it's it's a point where everything gets shredded, right? That is not true. It's a point where our theories break down. Classical theories show an infinite uh point to infinity, which is not theoretically possible. Nothing can be infinite mass. Right. Um, so that's why we have a singularity point that that actually uh says that okay, well here's where our theories break down, we need to look at
Speaker 5: different approaches. Um the other thing is um what are the alternative theories, right? There are okay, I'm closing down on time. Okay, I'll go fast. So there's quantum there's quantum gravity, uh which is uh studying uh well how how do um quantum mechanics behave once you get um inside of a black hole from a quantum mechanics perspective, uh which is something I can duck out on the hallway again. I'm running short on time. But this is the part that I really want to get into, uh Hawking radiation. Now uh according to physics, uh if I know uh most of people have heard that uh energy cannot be conserved
Speaker 5: energy is always conserved, it's never destroyed. Right? But from Hawking radiation, it's just everything that goes into a black hole, and if a black hole evaporates, well, where does the everything go? Okay, I'm still running short on time. Uh but where does it go? It just evaporates. And uh that's the most fascinating thing. I mean it's it's not something that we can directly test or directly say because uh trust me we don't have any probes in the air of black hole but theoretically uh it emits entangled particles which is uh uh very cold but still does not give us a definitive answer. So most uh I would end this talk by saying most approaches into understanding black holes are are from a quantum mechanics perspective
Speaker 5: and to understand well what happens to the fabric of um the universe as um a black hole unfolds Good, okay, alright so I still finish it in time
Speaker 6: Hello. So my name is Annette Lewis. I am a developer at Six Feet Up. I'm also a question asker. So let's get into this right away. What's hiding behind this button? Alright, so we're gonna start with something that seems like a simple thing. We need a registration button on the homepage. That seems straightforward. You know the site, you've got context, it seems easy. Update the template, add the button, add a link, done. Or are you? Maybe not. So what happens if registration is closed? One question has already changed the scope of your project. What looked like it was just simple and straightforward now has additional behavior associated with it So we've not even started really building yet, but there's more that you need to understand before you can actually accomplish your task.
Speaker 6: Another question, is a text always gonna say register? Well, someone says actually, marketing needs to be able to change that. Now it's not just a button. Someone needs control over your CTA. Can they change the text? Wait, can they change the link? How much control do they actually need? Again, that answer is going to change everything that you thought you knew. Is this only for the homepage? Now we need to think about reusability. Do we build something that's specific for the homepage? Well, they're like, well, for now, the homepage is the scope, but later it may be on other pages. You're thinking, does it need to be reusable now? Do I need to engineer that now?
Speaker 6: Does every version need to be the same? They're the individual or linked? Well, these are getting to be bigger and bigger questions. That simple button is nothing but simple So you gotta stop and think. What are we deciding without realizing it? In that first moment when you just said, I've got a button on the page, I'm gonna do it, you were starting maybe thinking about states. Now you're in display logic, reusability Or did you just jump straight to implementation and skip all of those steps? So as developers, we're often super eager to get solutions, especially if you envision something right away, you are ready to go and get that on the page. But After you hear that context and you start designing and you start thinking about the things, there's so many other questions. Your instinct may lead you to make assumptions and start answering questions that
Speaker 6: You might not actually have the answers to and you need to investigate more. So you've got a solution before you even understand the problem. So let's step back and ask a more open question What are you hoping this button will help your users do? This is a much bigger question. And Now we're not talking about implementation, where, how, what, we're talking about what is the actual goal, what are we trying to accomplish here? And now the whole conversation is going to change. That small button in the corner may technically satisfy the request and still do what you need, but if they say, oh, we want more people to find the registration page That small button might not have enough area to satisfy that at all. At that point, maybe you should have had a hero banner
Speaker 6: with a CTA, a block somewhere in a high traffic area, or even a card or featured content that repeats on pages Your solution completely changes. They asked you for a button, but that might not solve their problem. It might solve their problem, but that's where we don't want to make those assumptions. So We're actually going to start evaluating the solution that we had against the goal that we're trying to achieve. So something that I like to do is start thinking about questions, and these are some questions that Often help me start to uncover the right scope and what we're going for. There we go. What are we actually trying to accomplish? What is the goal of what we're trying to do? Who needs this and how would they use it? Who is the user that's going to be using this feature? What happens when things change?
Speaker 6: In the future, how are we going to manage the various versions and the iterations? How flexible does it need to be? Do we need to plan a lot of time now to make it reusable everywhere? Or do we just need it to work? And does the requested solution actually solve the problem? Will that little button actually help you Get people more on the registration page. So one of the most valuable things, of course, I've learned, especially in all of the years of consulting, working with different people, is To ask questions and actually get to the problem that they're trying to solve before we start defining the particulars of that solution And make it a point to understand what they're trying to accomplish. There's no automatic right answer, but it's important that we don't make decisions accidentally. So what is hiding behind this button?
Speaker 6: Maybe not much. Maybe quite a lot. But the right questions can take us from implementing a request to solving the real problem at hand. Thank you.
Speaker 7: All right. Who here has a dog, cat, or baby? Awesome. A lot of us. Who likes to send photos or videos of your pet or baby to friends and loved ones? Some of us, okay. Well I love to do it. Um so I'm gonna subject you all to it. Let me show you my dog Reese. He's a cocker spaniel that we got as a puppy a year ago and he's a lot of fun. Here he is patiently waiting for dinner, looking absolutely silly Now, I don't want to stand up here showing you all my photos and videos of him. That would take forever. Um some of you might want to see more, some of you might have wanted to see less, and I'm sorry.
Speaker 7: I had a similar feeling about sharing pictures with friends or family. I didn't want to spam them. Um and also I don't like using, you know, Instagram because it seems like 20% of that is People or people or pets that you follow, while the other 80% is ads or you know, what have you. Um So, like many of us here, I made a website about it. It is dailyreese. com, and you can see a new picture or video every day of Reese. So let's go over it. Let's see if I can do this right. Perfect. Um And this is all it is. It's a new picture or video
Speaker 7: every day. And wouldn't you believe it's made in Django? Well I'm logged in I can upload so if you all were to check you wouldn't see this little plus button. Um but I can. It'll auto-select the next available date that the uh photo or video is gonna appear. And it's also easy enough for my partner to use. Uh originally I like set up like I was like, oh, can you like download this SCP app and set it up to this web server? And she was just like, no. So Django to the rescue. When we upload, okay.
Speaker 7: So when we upload a a video file, um on my phone it's a. mov file. Um that fires a Okay, so here's the Django admin page. Um here's all the data things are gonna appear. Um when you upload a video file, it uses the new you know Django uh database backend task framework um to Convert videos to MP4 files, which is more web friendly. Um You can see the task results here. You can actually see. You can tell I just decided to do this presentation because I haven't really spent time looking at this bug
Speaker 7: where You know, it converts video. I think it has to do with system resources. It's a very small little web server, very cheap. Um but you know, just a huge stack trace here. And one day I'll solve it. But the uh What that looks like for people is that it's just an MOV file on the web. And the beauty of having um four or five Users is uh that there's not a lot of complaints. You know, when the when there's something wrong with the website, it's just like, you know, I might get one text every few months saying, hey This MOV file, or they don't say that. They just say, hey, this file isn't playing. And I'm like, oh, that friend has an Android, I I think, and it doesn't work. Once again, I'll get to that one day.
Speaker 7: And actually my favorite feature, I set this up on my partner's phone too, is a little iOS shortcut. I probably should have put it in the slides, but um Where from the photo you can click the little dog icon. Um iOS shortcuts is kind of like um a power user feature uh for iPhone. You can press the little dog icon and it'll just auto-upload it to a um an API endpoint set up with uh Django Ninja. It was like I was really pleased with how quickly I was able to do that. That was really cool. So shout out Django Ninja. I am so glad that I set this website up. Now I'd say my the main users are my sisters. uh my partner and you know a couple of friends and family um
Speaker 7: and they'll you know we'll text you know all the time about a cute photo or video um it's a lot of fun I also randomly get like a coworker commenting at it's like, oh wow, so some people actually, you know, still check this. That's fun. Um I no longer have to second guess myself on if I'm spamming someone. Um Do I think the future uh after Instagram is everyone setting up single-purpose websites like this? No. Uh but I've had a lot of fun doing it. Thank you.
Speaker 8: Sorry, not my computer, didn't bring mine, so hope I know which buttons to click here. Um hi, Mariana. I work at Linkin Loop. Um we're currently developing the Firefox website with Wagtail, which is pretty cool. I don't know if you've uh click around the Firefox website uh recently, but you know they have this new cool design, very pretty. Uh and that was implemented in Wagtail. Uh in turn that process, well the implementation started before the designs were finished, so a lot of those things were going on in parallel. Uh and I had a constant issue of
Speaker 8: I'm created a block and now this block needs a little tweak, has to change something. And I saw myself going like creating pages with all the possible variations of that block to make sure I didn't break anything. And so as my colleague Mark says, if I'm clicking a button, a butt a puppy dies. So I needed to find ways to not have to click buttons every time. So what I would do is I would create a page with all the possible variations copy that well so bactail pages are just json with content in that I would copy that content into a fixture and then run a command to create all those fixtures
Speaker 8: Uh so I could quickly look at all my blocks to see if I broke something, if they're still there, everything still renders as this supposed to be. And with time this evolved into a system where every block has a page. uh where we show up all the p possible variants, all the things that can go inside of it, uh and we have this collection of All the blocks, all the things that you can find on the CMS. So if you ever need to build a page, you have them all here. You can scroll to them, see what it looks like, edit the page. Uh and then we started to um make this the source of our tests.
Speaker 8: So we have tests oh no Uh what those uh fixtures look like, since that's the beauty of it, is just dictionaries and lists with Tax in them. That's what a fixture looks like. And then when I test those pages, I have those cool assertion methods that say All my headings are in the same the the right hierarchy. My buttons have all the things that they need to have. Uh the image shows up where it has to show up with the alt text, etc. etc. So this way we necessarily so you see there's nothing fancy here, just Python test, very simple, no library, not
Speaker 8: No fencing nothing. It's just a process where my documentation is necessarily tied to my tests. So Every time I change something on my blocks that will either break those nice pages or break my tests. So I make sure that everything is always in sync. I'm never losing data. I'm never like the next step will be to have visual regression tests on those pages. So we can make sure that a change didn't like Visual impact. So right now the Python tests only check that the elements are there and you know I'm not missing data anywhere. But you know the next step will be visual regressions where you can s
Speaker 8: take a picture, looks exactly the same, didn't break anything. And then our colleague built this nice uh explorer for the docs where you can like see the pages in line so you can get to navigate. uh through them. And it's all open source. So if you want like inspiration, I I think the nice thing about this is that we have a collection of things that make kind of every website kind of needs like an intro cards, a banner, an image with some text. Like they look different in each of the systems, but those kind of basic building blocks that kind of every website has. So if you're curious and you like want to reuse any of that, the project is very
Speaker 8: simple to run. So this is the GitHub on the readme there are links to the platform docs and like there are instructions to like that download the current uh database with all images And then you run the command to load the fixtures, uh, play around with it, and that's it. Any feedback, any ideas will be super welcome.
Speaker 6: Thank you.
Speaker 5: Taking photos at Django Kong is boring. What people do at DjangoCon They are talking about things that I don't understand, some differ uh funny sentences and numbers. They code while sitting they code On grass, they coat while standing. They go downstairs. They count on the floor. And on the floor. They code. And they are talking about talking about bootstrap, that I don't know what it is. They code outside. They code in a theater.
Speaker 5: Still they code and they type in different positions different places On the tables, under the table, by the pool, in a van In a red shirt with a red computer. And they code. But these people that gave me so many virtual hacks So many virtual hacks
Speaker 5: They do smile and laugh a lot. They do some weird stuff on stage. But they also talk about really important things on the stage that I can understand. They do some funny poses. They care about education of small children and want to pass the knowledge They hug, they hug a lot.
Speaker 5: They play games, they have fun with each other
Speaker 2: They have a lot of stickers on their laptop.
Speaker 5: They are still doing some weird stuff on stage. They self-care sprints, they sleep while coding. They sleep in different places than poses. Sometimes they look like uh Opera singers They cycle. Too much. They spend time together, they have fun, they are super stylish. They climb. They hike together. They swim together.
Speaker 5: They do sightseeing together. They eat together. They celebrate together. And they sacrificed them for organizing this event. They help each other a lot. These are great people. They juggle
Speaker 5: So this is the list of uh the conferences I've been to and the these photos that you see in right now were from uh these conferences. And if you want to see more, there is a album flicker with uh a lot more photos. And I wanna say just I need to see my notes. So I'm uh working as a photographer full-time and I've seen a couple different IT conferences This is the most welcoming one with really independent, real friendly and easy-going people. Taking photos here is part of my job, but I don't treat this as a normal assignment.
Speaker 5: If there is something interesting, I will try to be there and catch it. This is a place for some great thought-provoking topics for me as a non-technical person you are a bunch of badass mad max people independent trying to hack uh reality uh So yeah, call to action. Uh during sprints, I will be I would love to listen your stories about Django and how do you use Django and every day and at work as well and to record it, to to talk and to take some portraits. I'm here for you And thank you Jeff for involving me in 2016 and giving me this job and paraphrasing the uh
Speaker 5: the say uh I came for the client, but stayed for friends. Thank you.
Speaker 7: Can't believe I have to follow that. Um I was gonna talk about uh something I've been working on for almost two and a half years now and that's template diagnostics in your editor. Um so this is kind of an example of a wrong template tag. The auto escape tag only takes an on and an off argument. So this little error In the browser and I that kind of drives me crazy a little bit. I wish it was not just in the browser, but I kind of want it In my editor, sorry for the size of the picture. So there's a bunch of useful tools already.
Speaker 7: Framework, I believe, has a specific one for templates. There's many linters and formatters, but they don't really know anything about your project. And PyCharm um is great and it has a lot of this um but it's paid and uh closed and you know thank you Jeff Brains for sponsoring Django but um I I'm a NeoVim guy, so I kind of want something open that I can use in Neovim. And so that's why I started Django Language Server about two and a half years ago. And so my first thought for how do you um d bring dot
Speaker 8: diagnostics uh to the editor.
Speaker 7: I started with static Toml for all the tags and kind of came up with a little spec for it. Whether it was block tag, whether it had an int uh in tag, how many minimum or maximum arguments you could take and then the the choices Which worked, but it was kind of a pain because it meant you have to write your template tag, and then you also have to write the static Toml for just for validation
Speaker 8: if any new
Speaker 7: Versions of Django come out, you have to make sure that it's updated and it puts the onus on the user to, you know, if you have any of your custom tags. uh in your projects it meant you would have to write up too and it just some about it didn't sit right it worked um and so I took a step back and looked at kind of like the source code and what um temple tags kind of look like. And you if if you look at
Speaker 8: all uh
Speaker 7: everything that you need to like validate and generate um diagnostics are there. Um you can kinda tell that The the arguments comes from that bits variable. Um you can only have one because it can't um equal two and then can only be on or off. And then of course the big thing is the exceptions, the template syntax error and any exception that's raised in a template tag. Oh, that should have been later. Oh well. Anyway, so I that's what I did. I I uh Lost my mind uh parsing Python AST ASTs and extracting finding all the exceptions in a template tag, tracing it back to the
Speaker 7: any condition that led to that um exception and then um that's what I've been spending the past year uh doing so I was gonna show you a quick demo if I can get uh Uh, Linux, come on, there we go.
Speaker 8: Okay. Alright, so um here's that same uh auto escape uh example in Vim
Speaker 7: and it's not there. Oh no Um it would be there. We'll just skip the editor. Just pretend, you know, you saw the pictures earlier. It's got the nice red squig squigglies. Um but the the we it's also got a uh a check command uh um on the djls that's the binary that uh the server runs on um and so you can just run it and you get nice little um diagnostics right in your terminal um I also have uh pre-commit hooks, so um you can run that and it'll run it and so it'll run on any uh pre-commit um Any commit. And you can actually see I I had a spare one in there. Like we capture every you know, I've I've captured
Speaker 7: because I'm parsing the raw Python AST, um, it doesn't mean it means I can support pretty much any Template tag hopefully uh if I've thought of everything. Um but I also wanted to show uh as he creeps up on me. Um so I've got um a corpus of open source projects in the in the In the project, libraries are popular projects, one of them being Django packages. I run the test suite against all of them. There's about 80 of them, I think, or so.
Speaker 8: And so if I run it now, there's no there's no errors because Django
Speaker 7: Packages. org is an excellently uh maintained project, but if I run this thing it's gonna delete all the end if or the uh end block no end if um blocks and if I run that again You can see uh it found ten errors and pretty quick. Um like uh
Speaker 8: my status line usually says you know like
Speaker 7: took two seconds, too took five seconds. It couldn't even register it and went so quick. So anyway Uh we get back to we've got VS Code Z Neo Vim Sublime. Um there's the pre-commit hook. if you want to use it and all the features I don't have time to go through. We have a lot of feet I've built a lot of features of the language server protocol. I'm missing Still a lot, so uh probably work to do. And I'll be spurning on it if you want to come and try it out in your project. I'd love to get issues out of it. I have
Speaker 8: done it in my on my machine. It works on my machine. I'd love to see if it works in your machine and your projects.
Speaker 7: So I'll be here the next two days. Thanks. We'll just do uh Adam Power.
August 24 - 28, 2026
Chicago, Illinois, USA
Note: We understand that names change, people change, and bodies change. We respect each individual's journey and privacy. If you have any concerns about a video or need us to remove content, please don't hesitate to contact us. We will handle your request with care and promptly address any issues.
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026
Published October 26, 2026