Keeping Django relevant for the next 20 years - Thibaud Colas

Speaker 1: All right, uh hi, I'm Thibault Pronounce Eim. I'm a dev, and I'm also the president of the Django Software Foundation. Which is a pretty cool sounding title if you ask me. I don't think you know what it means, but uh I thought I'd share, I guess how I got here and how I think about the future of Django because it's a pretty big deal that 's all of us, uh President Arnold. Um just before I get started a quick thank you to my partner Deborah. looking after our kids for I get to be here so thank you for our colleagues. Yeah Django, um I 'd love to start with a question for the room Does anyone know how old Changle

Speaker 1: is as a framework? I'm sure some of you know the answer Yeah, uh

Speaker 2: nearly twenty.

Speaker 1: Yeah, nearly twenty. So I think twenty in July is quite a big age. Um who here has used Django for let's say at least a year? Five years? Okay, five years great. Ten years? I need like 10 years checkable users? Remember the one little days? No? Okay, great. 15 years? Oh wow, John, wow.

Speaker 3: Not strictly a user, but

Speaker 1: it counts. I'll take it. Yeah, it's been around since forever. Ruby only has been us by a few months, but yeah, cool. Yeah, the web framework for Christianists with deadlines quite a quite a big uh headline that I think is still very relevant. Something people might not know about Django. New Discord you might know already, but it's run by foundations since 2008. Back then it was super new for open source projects to be run by foundations other than just random contributors or companies. We were one of the first ones in the web framework space to show. And an even bigger deal is Django has no VDFs. Since 2014.

Speaker 1: Carlo Van Rosen for Python is the Python PDFL, the dictator for light he used to call the shots. on web dog in and out of ice other language. And Django hasn't had that for ten years. And definitely, since it's a community run project with a foundation, it is quite tricky for us to make decisions on XIT uh how the thream watch people without this kind of singular vision. So we can talk um we can just want to know space quite spread for three more back then to have made that choice And uh yeah, it's used all over the place. Hundred K is the number of projects I'm running with is definitely hundreds of hundreds , millions of apps using Django and even more people using Django

Speaker 1: projects. I can always use it on websites and apps. Um yeah. Great. Talk about Django anymore. Me personally, I've been using Django for about ten years now, here and there. um specifically on the code Django and a bit of the community things like Django Girls Plus. And I joined the DSFOS, so the foundation that was on the The projects in late twenty twenty three or a two year term over twenty four and twenty twenty five And I'm the president as of a few weeks ago. Um I don't know quite well what it means. I have no idea

Speaker 1: where to go further after having been the president. Yeah, Adam, you've been around in the Django space for a while, and you tell me what makes sense to you after really you have a little in all these data as well, or you know, I have no idea. I'm just making it up as I go along and figure it out. Uh so yeah if you want to do if you want to become the president of Django, uh that sounds cool to you, this is kind of what I would recommend. The the main purpose of today's talk, I guess, is to convince people who are in this space right there to move forward to the And again, I can I can you know maybe keep you the other online first. So the thing you can do today is like save astronom, help

Speaker 1: us as the latest version of Django. So Django 5. 2 Coming up in April. It will only come up come out in time and vote free if we have people from the community testing it, which means people in this school will take the time whether it's in continuous integration or just you know installing on your local copy of the project and buying out. Yeah so I guess to move from this kind of user developer mindset to contribute our mindset. I like to think of why first of all do people bother? Why why do you take the time? And I guess in my case I think of kind of impacts that can have. So for me personally I want to illustrate this a bit what it means to try and have an impact by looking at the accessibility scores

Speaker 1: of different web frameworks. So we don't necessarily score like the framework itself, it's how the spice deals with this other framework and with Geo 's spice. on average looking all the expected case. So Jengo's rights like in the middle-ish. Those to be in the middle here are really where it sets is uh it's It's probably a lower than the average website out there. The average of site is at 84, whatever that means Django is 82. So why are you available with Django? is to get us from like place number two or three to number one spots on accessibility. And yeah, that takes you know lots of lots of telephones from lots of people. And even if you don't want to do like

Speaker 1: this kind of specific R S progress, just helping the framework that you use as a dev that you get for free. You can convince your boss to sponsor the project also a very solid thing to do. Um yeah, so just like again on my motivation is This is the score over time. So basically we have those numbers out there. We know how different stood compare of across the whole web framework ecosystem and Python specifically, and I want to see your As everyone else does, but I think it's good to keep in mind those kinds of underlying motivations. Um and yeah, why I guess I became the president's The board member in particular is uh this graph right there. I I don't know who in the room would have seen this before.

Speaker 1: If you have probably It pumps you there 's night nine if you haven't stressed carbon emissions over time. So if you have climate change And what I came to realize is uh as a dev, our industry definitely has a part to play in this. I'm sure maybe transportation, maybe publicities and so on, and so forth. But even tools like Django at the scale of 500,000 projects, you can imagine each and every one of those projects adding a server behind it. even two, maybe three servers, loads of people loading those projects in their browsers to use them. Any kind of change we can do at the scale of half a million projects that has that slime go down. Um and again like Django

Speaker 1: is is just a web framework there are many people problems out there like transportation, um, air shuttle, but as a dev, I feel like I can have a small part in in this helping scale framework to us, things that have lower lower food rates And yeah, so again that's what helped me into the whole like police trying to get the community a bit more. And um when you are in that kind of place, in a kind of all -member mindset It is up to have to wonder about different things than your own custom dishes. That's why I'm talking here today to talk about the overall health of the projects and uh challenges that we have. Um this this talk used to be called um those

Speaker 1: existential threats And I was already asked by my other board members to not see that term , maybe it's too big with the world. So yeah, it's just pandages. Um so this this track right there again is another chart that's quite hard to interface. We're comparing lots of frameworks and looking at the frameworks websites how much traffic it gets from people. The idea being that if a framework's website gets lots of usage, only it means the framework is too large. So the lower on the charts, the more popular the framework is. And at the very bottom right there, you have what I call the WC club, the boring club of tech that's not gonna go anywhere not gonna disappear ever.

Speaker 1: So Django's the bigger line right here. Django, no otherwise ARP existential threats, it's gonna be there in 30 years time, no question. uh lots of sites have their queries, lots of devs have their queries, built upon it, Python's super popular, Django is not going anywhere. The only question I guess is how active we all want to be as far as making it more popular and maybe having it adopt more modern aspects of web dev I guess just like whether we're going slightly up or slightly. But yeah, no question that if you pick up today, one of these big strengths, you know you can have grades every version every nine months, no problem But yeah, looking at other frameworks, it's definitely interesting to look at the trends.

Speaker 1: So there is a few lines here over to the right in particular that built down classified What that means is new frameworks out there getting really popular really quickly. So that one right there, that's the Astro web framework. It's kind of uh another tech on the Netflix of the world. So Netflix right here quite fast in 2020 and it's more or less stable. Astral is going like this So that's the kind of thing where as a board member I want to think of what I think that we are doing that we should consider as well. There's a there's a catch here which is that the next JS on the past role the world they are by uh venture capital funding

Speaker 1: I don't know who knows, but NSA is still on the order of a hundred million, maybe a hundred and fifty now of enter capital funding. to make the framework as cool as possible and as tight as possible in particular. So keep that in mind when you compare your web frameworks, which ones have something to say to you and which ones So Jango in particular are marketing budget for this year. Can you guess what it is? Zero, yes. We have a zero zero the lab marketing budget. It's all going to be powerless. Uh yeah, we're sold next JS. Uh probably ends up minutes. And yeah, so again, just like one more line to keep in mind. This one right there is HTMX, just like going straight down.

Speaker 1: And like this, that means HTMX is up right now. So that's where I want to look. think okay those ones what can we learn from them what can we learn about the marketing from them I don't know here who's on social media HGMS memes are super hot So if you're not contributing by testing Django's latest alpha release, definitely make some memes about Django. I have all our social media accounts, I'll happily repost them just to keep the high up So yeah, just one last thing on this. Um the charts Tells us what people are using, which frame what they are using, doesn't tell us whether they have them or not. So that's what I want to look at next. Which things of Django should we improve for people who are already happy using it, but could be happier

Speaker 1: So now return to another dataset which is the Sype Overflow Developer Survey. I don't know here whose ANAS or ISPOI, but it gets like tens of thousands or something. Every year is probably one of the most comprehensive surveys across the whole debt ecosystem. We're looking at numbers or generals of the IC now over the last six or so years of the survey And the numbers, I guess, if you look just at the ones that I highlighted here, they're not so good. So people who want to start using general has come from 9%. in twenty eighteen to three percent now. That's uh to me it sounds like a problem. So that means that there are fewer people out there who haven't used Django yet in one individual. despite Python itself becoming more popular as a language.

Speaker 1: So that's the kind of thing where I want us to question why that might be and think of whether it's marketing. There is documentation, there is features of the framework are to reverse that trend. And yet people who want to stop using Django. So LC18, 41% of existing Django users wanted to stop using it That numbers don't happen. So also the wrong direction. So it's quite a big picture, but I think it's worth saying as well. This survey, there's only a little number of frameworks on there Those are sort of effects here where the frameworks that I'm going to call I cut new sheets, they're not even on the survey at all. The financial for the incumbent frameworks have that kind of change. But yeah, if we try to adjust for all those kinds of underlying trends, that's the kind of numbers I've arrived at.

Speaker 1: So People who work with Django that goes up. People who want to work with Django actually goes up compared to the other frameworks in the survey so again like overall it goes down but if you look at all the frameworks they're all going down this kind of line shine but it means it's not so bad necessarily and In past you have to want to see look people who are stopped link Jet Go are not those actually are good on this compared to other frameworks So yeah, again like not necessarily a problem there, but just something that I think as people move from the user slash developer mindset to contributor mindset It's really cool to be aware of those trends and of uh yeah what what moves the needle for

Speaker 1: those projects This this chart I think I might skip over. So again, it's just looking at these A numbers as a as a timeline and just to illustrate. This is not so bad. But yeah, one more chart to unpack. That's uh with that solver. Um I guess the the takeaway thing for me here is just that people who use Django based securities uh more or less, that's kind of what the data shows for us. We just need to find convince more new Python users to do that well. So yeah, what do we do about this? How do we need it's really helpful to look at other Python frameworks to see see what's happening and here we're switching to the Python developer survey.

Speaker 1: Oh this is gonna be hard to read I don't know as the comparing uh user frameworks of people in the data science space and the people with the web dev space. So in all cases there's lots of lots more web dev people within those frameworks, which makes sense. What I think is interesting here is the difference in framework usage between data science course and the web death course It's really small for class. That means amongst the people, clask is reasonably as popular as it is for web devs. I say yeah, bigger difference. Django much bigger difference. So to me right there there's a question, why aren't data science people within Django? Maybe they don't need the RM, maybe they don't need the anime, that's completely fine Maybe we need to tell a better story to our users so they feel more comfortable using Django, even if they only need, say,

Speaker 1: twenty percent of the features Maybe if people love like single file web frameworks so much, we should encourage people to generally single file models. Which is definitely supported, something called micro Django. Um, I don't think there's any like underlying overrice Django to have to make that work. It's just configuring it definitely for it to work. Maybe we need to focus on the developer experience improvement. those frameworks and make them happen for the wisdom. And um yeah maybe maybe we need to have an SROM typing in Python as well Who here likes types in their Python code? Can you show your hand if you like types? One hand doesn't like types, lots of hands that do.

Speaker 1: Who here actually types their project? Okay, yeah, that's the kind of shine that's yeah. I think I don't think necessary Django has to adopt typing, but we have to have a good extra because the people who use FAST API, well let me tell you they talk about types all the time and how much you love them. Yeah, there's no ninja for API users. Yeah, for the typing to Django that people tend to like as well. But yeah, I just think we have to have a position on this and think of other things with Django overall. Um Google uses RAF, uh RAF the internal informator. Yeah, a few people. Um it's gonna be adding a type checker in one of their future releases.

Speaker 1: They have an only project called Red Not to look at student tests into Rough Uh let me tell you, with real popularity rising from month to month, year to year, then adding a type checker into a tool that people already use anyway, it's gonna be a big deal for Python and go as well. So again, the discovery system they comment on the online reasons and I already mentioned I think and the other aspects of people tend to mention the performance and the developer ergonomics. So definitely things where performance are already not that big of a difference. That requires people frameworks, developer ergonomics

Speaker 1: Types between systems and within systems. I'll show on the technical government Yeah, so it's not all data driven. Reason I'm in this room just to give a little bit of data study on your old on maybe pain points you have to dangle and to help us shape our direction As the president I already gets to tell other people what we should do with the framework technically. People agree in this room as well. They have an unfortunate problem with direction. the technical aspects of the project, but I can definitely really add back to the right people. And yeah, in my mind, that's one of the biggest things we need to do more is talking to the users. And uh yeah, there's talking in the room, but there's also just reading people's meaning comments on the internet.

Speaker 1: Um which you know if you get to manage Django 's social media accounts you can spend as much time as you like reading people's upstakes on why Django is dead or why it's boring or you know how much hype there is in Jay's world of excitement. So yeah, I just want to show a few highlights up there. I want to be crystal clear. No matter how much fun I make of this, it's amazing for us to get that kind of tough feedback in our faces. And the ones I picked here are definitely very eloquent arguments that people really have spent time providing us with constructive feedback. So life and the pace of evolution, I think they mean just like we're moving too slow. I want to hear that. I don't think it's necessary true, but I want to make sure that we take

Speaker 1: into account when we market the framework. And uh yeah, uh one of the people that replied to this they said, oh, there are no fake corporations with massive money back in Django. And that's kind of true to be honest. The people that tend to pack Django, there's very few of them compared to how many users we get And the donations is definitely not millions of metric capital dollars, it's more like tens of thousands. And all of that gets paid back to changed development. Um we have John in this room. John 's company is one of our corporate sponsors. Thank you, John, for being about this. So we need more more turns and we need more

Speaker 1: possibility talents to help us uh you know financially have a s have a future for Django. I don't know here who knows this, but uh the Django fellows let's say mentioned in his talk, the people who originally work on Django, that's uh where their money comes, the their salaries come from is composed also for the projects So yeah, Changel is a foundation, it's a nonprofit, it's basically seven or members who know as much about nonprofits as your local uh village church in the UK you really didn't come to this extra hard extra teach learning foundations and non-profits. So that's kind of why the place and the marketing might not be so good. Uh it's not because of our willingness to

Speaker 1: play with stick with purpose of Django. I have no idea the purpose of Django is a value take, but definitely, you know, in my mind. The reason why we're here, why we might be lacking in innovation at base is competing in those frameworks with the marketing budgets in the millions. And yeah, definitely you know if you want a fast pace of innovation. The JS ecosystem, I I love JS as a language. Uh just the other day I spent like ten hours just switching from one tool to another. With multicipants, again I I love React as a library to the React ecosystem, like how much manuals I'm going from 16 to 17 to 18 to 19 Oh boy.

Speaker 1: Yeah, just infinite sources of excitement and innovation. And again, I say that as someone who loves reactive stuff as I So yeah, existential threats and I guess including through those. You want to keep the diagnosis? We want to retame them. And Django having been around for so long, having so much code that's as shown, is using places don't necessarily expect It's hard to do the chainable, there's no question around this. And that's why we need programs like the space, as Save mentioned, as a program to help us on all the contributions Yeah, retention is how I have a link here. People are kind of interested in the data behind this.

Speaker 1: I have a link that shows you a full report of uh statistics on commits and pull requests in in general over time and one of the graphs on there shows you how many contributors we retain over the last year Two years, three years, five, seven years, ten years, and the numbers are being much smaller than people might expect. So it's a real challenge for us. And yeah, governance as well. So I mentioned Django not having a benevolent data for life. So definitely maintaining this vision of what Django is over time and the technical direction interpret that you have to be a bit of how the industry holding that is a challenge and yeah the steering council who really is a member of that's kind of a group of steers that are in charge of helping the community

Speaker 1: overall with consensus or otherwise deciding how negotiable really tough tough task but also for the rest of it. So yeah, opportunities. I thought I showcased a few specific ones and I thought I showcased one in particular that I got started um about a year ago now, which is the opportunity for Django to have a roadmap. So what we're looking at here is a list of potential changes to the projects or the community around this that have gotten thumbs up for enough people to make an interesting list to look at. This is not official by any This isn't something I did as a president or even as a board member, but that's exactly the type of thing that I think is important for us to put together as a community.

Speaker 1: And um if we look from the top like five to ten, this is gonna happen this year. This happened last year, I wouldn't love for this to happen very soon. And um tell us how to go about this, wanna make it happen. Let's

Speaker 2: transfer the repo, then

Speaker 1: yeah. The built-in API framework. I know there's lots of conversations about this. Do we have this now? Not not yet? Okay. It's good. Uh oh template languages. I could I could spend time on that. Um yeah. I I'd love for us to move to DJ for projects. Um I won't know, but Yeah, open everyone has gonna I would have it done. Um Okay then. Okay, yes. I hope you all have questions almost done. So yes, the roadmap we need one

Speaker 1: individual for our package ecosystem. There's lots of discussions about should we add this to Django, should we add that? Should we have a better API framework? Should we have a better language? Should we have Django upgrades? Our answer tends to be pleased packaged ecosystem, which is I think now 18,000 manual packages, it is hard to navigate all of that, but it is definitely way more room for us to innovate collectively in there But I think the foundation needs to do in Pascal though is having a vision for how we do that. So how do that do you think the day-to-day combine some of those 18,000 packages into their project as a non-trivial task. A more professional foundation, yeah. Again, I I'm a volunteer.

Speaker 1: I marketing is not my forte, so we might just need to hire more people who are paid to do this for a living. That needs more funding though. something to keep in mind. And yeah, more discussion to the community. So again I want to make clear the the main internet comments I I um to the leading ads uh it's great for us to have them. I think it's just important to have a kind of back and forth. and also maybe to have more formal ways to give the feedback. So the surveys in particular is a couple of what you need to take. time to complete those surveys to use the results to inform our direction. Yeah, that's me if you use Django. Um um say hi and continuous speaker and uh yeah uh then tell me how to use Django

Speaker 1: and also tell me how to convince your boss to sponsor the project

Speaker 4: On types, yeah I was interested in most people seem enthusiastic about types. Um my experience to be yeah, times on battery project. It's it's all you need to be really good at them. But I am able to kill my um it's I I feel like this comment I made. It feels to me like It's really important to have timing on the interfaces where you like if you're going to send example on a team, who's another service there, but then internally within the project, it feels when you have time to just kind of resistance. Because that makes sense as a distinction.

Speaker 1: Yeah, 100%. I don't get to hold it any more like a particular opinion. Anyway , I can tell you my opinion. Um my opinion is it makes lots of sense for library, like Django, because lots of people spend loads of time looking at this So I'm trying to figure out how to do this. So whether you like like the I guess correctness checking aspect of types, I'm not sure that's even relevant because just as far as documentation, it goes a long way to be able to say how distinct. based on that time. At the scale of Django, you know it might take time to write those types, but also you have to literally millions of types are of the them and they don't know how to do it, you know, it's just like if they want to use it, they have to.

Speaker 4: But then to be a full test types.

Speaker 1: No, so then I guess it's up to every project to decide. So I guess if we work with other types in Dango, we would want Lego 's have things So every resistance uh excited. I haven't mentioned I haven't interested the whole type without talking about AI, but I think there's something to say about types and AI as well. Any questions about AI then no? Sorry.

Speaker 4: Sorry, this this isn't AI Um

Speaker 5: so any project that people are actually using is gonna have like features of a lot of control or computer to be wallet and feature Okay, cool. So uh so yeah, that's that's natural, you can have sort of feature requests. But um Ginger 2 as a feature isn't a thing that's stopping or maybe it's about considerable Django. I'm curious how do you identify what features or aspects of Django stopping people from using it or go to putting up the rather than just what's the most popular missing feature?

Speaker 1: The question is. So um Little little caveat here as um said from my bonus apologies here. I also do this kind of thing as a day job on the actual CMS, which is a good one. So I can tell you for Wikipedia is really simple. You just spend lots of time talking to people and asking them what's made you choose this We should what other options are you considering or are you dealing breakers? And uh people might not tell you when it's just their personal size, but when it's uh 500 company, they'll tell you exactly what specific breakers and not even for the private bit, but just UK governments adopting a framework deciding between uh say Django and SJS

Speaker 1: projects, U here, but here that might be good. the pain points. So in my mind, there's two aspects to this question. There is being around having people around in a couple discussions. Just have to get a sense of the bigger picture. And um then there's actually having better marketing to be honest. Where people say, Oh, we want like this shiny new thing when actually what Muslim for them maybe the framework being super fully stable. They literally just say like, oh, I only have to upgrade the measure number in my defensive file, and what is this? I don't know many frameworks of the side of table that can make that happen

Speaker 1: So yeah, big around discussions with people compare frameworks, having people that sign maybe Jangle plus variable and asking them which aspects of variable you want to steal, which things are in your killer features there, then you are seeing a for this and understand what the death from the framework. Any more questions?

Speaker 6: Hey, um yeah, thanks so much for the um talk. Really interesting. So let's see it from like a business competitor perspective, not seeing another girl There was one graph you showed I think was like all 10 processor frameworks and I noticed there is a it was the one we talked about at entry capital document That's the one. I'm curious what happened till mid-2023.

Speaker 1: So when I got there

Speaker 6: Yeah.

Speaker 1: Probably nothing. It's the challenge. The bigger thing is uh having this kind of data from getting frameworks is really tricky The numbers I have to look at are the download counts and download counts from ecosystem to ecosystem they have very different stories because they depend a lot on how much CIA adoption area, stuff like that. So this basically has set called Trenko. It essentially has all of the websites in the world, or at least 20 million of them. And just having this kind of data is hard to manage. So yeah, this is probably just a belief that ACS has whether anything actually will change. in usage of the framework. But yeah, look beyond the one list, like look longer term, the trend lines, and you can see some trend lines being really fast

Speaker 1: and some being so definitely key

Speaker 6: I'm just saying it's not

Speaker 5: anymore All right, well we can talk a deeper workers. Thank you. Thank you again.