Transcript: Django on the Med

I'm Will Vincent with Carlton Gibson.

Hey Carlton.

Hello, Will.

We did a sneak peek one for the developer survey, but this is the first start of the fall season.

Yeah.

It still took us, what, 20 minutes to figure out the logistics of the technology?

Yeah.

Yeah.

Well, you look tanned and rested.

Tell us where have you been?

It's growing like a cancer, it's doubling.

That would be in the North Sea.

Yeah, yeah, yeah.

Anyway, let's let's leave that one, kids.

Okay, so 25.

25, yeah.

We had 15 last time and we had 25 this time.

Um, so obviously growing.

I think the um last time was a bit of a punt.

It was an untried thing, and so the very only the the very brave could make it.

Um this time it was a slightly more of a known commodity, and so um actually we we had a capacity limit.

We were we were we weren't sold out because we didn't sell anything, but um, it was free to attend.

But the tickets were sold out a month more over a month in advance.

So that's really uh that's really nice.

Yeah, and then um, and you've done your part, but I've seen Paolo in person at PyCon US, EuroPython, DjangoCon US.

He's you know relentless with letting people know about this.

So um that's exciting.

I mean we'll we'll get to there's some stuff that you worked on that came out of it, and we'll sort of do a deep dive for people who just

But uh, I'm sure we've done more than just that.

It was last week.

What was the rest of the the things?

Um I have some notes here.

Uh we'll we'll skip over some spicy takes.

You know, I'll just say glad we're not the Rails community.

Let's leave it at that.

Um so DjangoCon US um in Chicago, that was last month.

Um great, so nice to see people.

Um I miss DjangoCon Europe in Athens.

Uh so got to go to the PyCon US and EuroPython.

So it's nice to be in like a cozy Django little world.

I gave my um deployment speech, which I've given this year, um, super focused on Django.

So that I think that went well.

Um oh, uh and just for people who don't know, um, because you're not so subscribed to the Django newsletter, uh Django has finished, I think, an executive director search, or that's still ongoing.

There's been some tech governance changes, right?

The steering council, as well as versioning.

So one was it we'll go backwards because um debt 20 was approved before debt 19.

Deb 19 took longer.

So debt 20 was the move to the annual release cycles, um, which

is so good, so good.

Um the goal there was twofold, really.

Um one was to bring uh Django better in line with the Python release cycles.

So since Python a few years ago went um annual release cycles, we've the Django's old Django's old release cycle was leading us in leading us at sort of well out of sync with that.

So Django 4.2, we just went out of life.

I think we're still supporting Django 3.8, which had been end of life for 18 months before Django.

And it we don't want to be supporting end of life um Python's it is a start.

But um so we want to get better and um a better cadence in in line with the Python's release cycle, and also to eliminate this gap between LTS um releases.

So we've had every third release being an LTS, which has three years of extended support, well that that would mean that people would stay on the old LTS.

And what's going to happen from Django 2028, which will be the first version of Django under the new system, every single version will get that same three years of support.

And the idea being that if you want you want to be on a long-term support version, well, that's fine.

You can still update and get the fixes that come in on the in the new version, um, as well as the features.

It's uh you know, if people who are hanging around on the old LTS, it's often about the the the fixes rather than the features.

Right.

But right.

We wanted to have carrot and stick.

No, it made it makes so much sense.

I mean, and the old system made total sense when it was first done.

Um but this is I mean, even PyCharm, my day job, you know, we have an annual release cycle and it just makes so much sense.

And you know, to the extent there's humans making content, it makes it easier to say this is not out of date.

Um that's me sliding in.

So last night, if I seem extra tired, I finished, finished, finished uh Django for Beginners updates.

So for 6-1, uh maybe I won't subject people to a whole episode on it, but I've actually gonna have one book instead of three.

So I moved the API's book into a chapter or two, I slimmed down the uh professionals book into a chapter or two, and I'm just gonna be so I won't get rid of the old versions, they'll still be accessible to people who want them, but one book up to date, and then I guess 2028 will be the next update.

So I'll I'll subject people when I have a cover.

It's not real until I have a cover.

You know, for me to be like, oh, all the text is updated, all the code is updated, it's like I want to see the cover.

it as um six point X?

So for the six point two LTS to come up and then

But it's um I I should be with one book, I should be able to just bump it every eight months.

Um I I it it's not as simple as bumping it.

Um people don't want to hear me lament being a technical book author, book author, but it's like it's the least of it to do it.

But it should be much more manageable.

Um and so I'm excited about that because there's a ton of new stuff with um you know mailers and then the fetch modes and um and at the same time the professionals book had a lot of is the one that's a bit old, um, had a lot of Docker stuff.

Like I feel like that's that's a nut that's hard to crack.

So um even though it's Django for beginners, it's really takes you into some professional territory.

Um, and it's just um yeah, anyways, I'm doing the plug for the book that wasn't gonna do, but I'm just excited because I stayed up last night and finally finished it.

Um and certain eight tooling can help us to produce, you know, the exact Docker file we need this week.

Um but the understanding of the general principles around deployment, that's that's something that you know a human can bring to the table at a higher level, I think.

Um and also I found out this summer, um, Sarah Abdermain is one of the two authors updating Django by example, which is the other ongoing book.

Um so when they have an update, we'll bring her on and we can have a book off.

And uh

I don't know who it is yet.

She was writing a uh chapter on um Django Tasks, who's bemoaning to me, not giving them giving her enough pages to do it.

I need fellow tech authors to like Okay,

support group on the podcast one F to see.

So DEP 10 uh was the one which when we moved from um uh the core the the old core developer mailing list, the private core developer mailing list, and um retired the core the core developer label and the old team there, and we had uh DEP 10, which was super.

It did it got us out of the um the history, but it turned it turned out to be quite um legalistic and difficult to work with.

Um and so the idea in DEP 19 is to very much loosen the the the feels and have uh something that's slightly more human, slightly more um fitted to the scale of organization that Django really is.

We're a very small and tight community, and sometimes trying to legislate for uh yeah situations which were uh uh far removed from the norm, didn't leave the the day-to-day you working with Deb 10 and then Deb 12 where we tried to clarify it.

Um it wasn't that easy.

Um and so the big the Tim Schilling has worked on that very heavily.

Um so full credit to him.

Uh he really has pushed that effort.

And I think one of the bits we're most pleased with is uh the section where it describes some of the candidate features that you might meet.

It's kind of like a, you know, you might have two or three of these of these qualities and you'd be qualified.

And if you read it, it basically anybody who's engaged in the Django community should be qualified and should at least have written down that you are qualified.

Whereas previously it was very much like you must have five years of serious contribution, yeah, which makes it sound as if only the old grey beards can be involved, and that's not what we want.

Um speaking of which

So maybe I'll even check

as the weekend.

It's been a hot summer, and I was at the point where I said, No, it's gotta go for a while.

Um so maybe I pull it back.

what passes

No it's a nice hobby.

So I can I can get it back.

It's way less work to be clean shaven ish.

That's that's the you know you see someone with the hair and it's like oh the maintenance I can't anyways um I do want to say so Jane going the med you had uh some sponsors should we shout them out

on Italia who came on board as community sponsors as well or community is you know community sponsors or community uh partners uh perhaps shall we say um we did try and get a few visa applications for people from outside the the the who didn't have visas that turns out to be very difficult um pythonitalia they bent over backwards to try and help us there um but I think for a small organisation like Django and the or for a small event like Django on the Med it's really hard to see how that's feasible um with the amount of work it

takes um so maybe the DSF can step forward and find a way of doing that in the future I don't know that's something that the board can discuss but it's it really isn't simple because it it's not the case that an individual can just sponsor somebody to to come you can't it i it's it's quite it's a lot more to it than we than I think we've seen before.

But I did want to say so Python Pescara the local group I think also

to that when he gets a chance to put it up

yeah so Paolo has a family friend who came and did photos like you know missing Bartek so we had Alicia and um and then um she was doing videos and she's putting those together we've got Django Med's Instagram account now of all things so do follow that if you're on IG people this is so far beyond like you know what we what we do normally so Paolo really has pushed the pushed the limit far out so we're never gonna be able to compete with this in the

future I

Um Joe Zamet who's uh who lives there he's gonna be on the ground there um and put together so we'll be in Malta probably in the second week of October um which is just

And then I saw um well

Oh yes you're right it is I appreciate the joke Carlton come on and then there's talk of Django off the med, you know, in Cambridge and other places.

So we'll see how it

ter that we spin off um you know other events elsewhere and then you know some people can travel sure so we had Lillian um we had Simon Sherette we had um uh Jacob Wolves they all came a long way but um it the the vast majority were much more local to Europe you know Austria Italy Spain you know these kind of striking distance type um type locations and that's I think a much more sustainable it's going to be a much more sustainable model but that requires that we have people take the bat on

and and run one in Britain or run one in America or run one you know we had various people who like oh we could do one in the Alps we could do you know so well

Well because DjangoCon Europe needs a backdrop and the pictures of Innsbruck with the mountains right behind it it looks very looks very nice.

like you know there's been put um blog posts already a couple of blog posts already published and the and the key point that keeps getting made is that this these these um advances they wouldn't be possible without the time spent in person it really is a different environment than online um so matthias uh whose surname has totally escaped me um fine cms person ah matthias and jo and then joe um who's coding joe they they're both working on the front end they've both been working on import maps they've both been working on um you know asset handling but com

ing from different approaches and then they get together and they sit around a table and within an hour they've realized that actually their approaches could be compatible and if they just did it this way then they could come up with something which satisfies both their needs.

Now that but that conversation has been going for literally years on the forum if you go on the forum and you you look up for about the form media and and and and related tickets around there it's like two, three years of chipping away and bubbling and and whatnot versus an hour in the room together.

It's another I mean because part of the issue you know sprints are in some ways the best part of DjangoCon conferences, but you're already four or five days in, right?

So you're already a little fatigued and not everyone can you know extend.

So to have so I assume that's you know in person but also like you're kind of like okay I'm gonna see Carlton like let's let's hash this out and then to your point it's usually in yeah in person you can get way more done.

I mean that was yeah we we we've lamented it I mean we we did this podcast in part because without the Django cons for you know during COVID not that it falls apart but it's a lot harder.

As digital as we all are like you do need that in person.

high performance team who are working on the um subatomic package and um nested nested atomic and we were able to um have probably half an hour 40 minutes where it explained the the problems and the issues and we went through the foot guns and just brainstorming back and forth and by the end of that I was able to get a com that morning I was able to put a really nice comment on the on the um issue which then Caleb Hattinger who was another sprint attendee picked up and then on the second day when I was busy um reviewing a

um uh another H Ink PR which has been on my backlist for six months he was able to make progress on the basis of the comment and then on the third day we were able to work together and I think you know within a few weeks we will have something so I'm at I'm very excited that we will get this into Jango 6.2 now but that first of all that big session with the the you know half dozen people working together where we we were able to to reach an understanding which otherwise wouldn't be there.

Then him to be able to pick up that on the basis of it then we that that work just literally isn't rec reprobical in just an open source environment an online environment um so that's phenomenal and then you know I could talk for days and days about the other the other things are going on.

I don't want to minimize

So Mark Smith and I did um bootstrapped a package based on so Hinnick put together a package called um uh but BGT uh for background threads, I guess it stands for.

But um I think we we we think we should pronounce it Django beer Gut because it's BGT.

Um so but we we put together a Django version of that.

So that um that enables you to run um supervised uh background threads um uh with exponential back off and retries and all the all these things you want.

Plus, it's got a uh leadership um mechanism where um if you run it in multiple processes, you can set for one of them to take a lock via a lease um table in that the ORM provides, and then and that would only be run once.

So you might say want the report, the Monday morning report to the CEO.

It's really important that that only goes out once.

You don't want like 10 of those arriving on the CEO CEO's desk at once.

So you'd have a leader for that, and that that can only happen the one time.

Um, but that's a very exciting um additional tool in the toolbox, the exponential back-off, the leadership, that that kind of those those kind of features are very hard to write, and we're leveraging um Henix expertise there.

But that um that can work then with any database that um the the ORM supports, so you know, SQLite or uh MySQL or PostgreSQL or Oracle or you know, whatever.

Um so very excited about that.

There's a there's a channels wake up.

Um so wake up, so you normally um by default it wakes up on an interval, so you know it's like a cron job, you know, you have it at um every every hour, say, clear out your sessions.

But also um there's a channels wake up where you can subscribe to a channels group and then it will wake up on an event to the channels group.

That's in the code base, but it's not documented or tested yet because we didn't quite get to it, but that will be there shortly.

And it's yeah, it's gonna be a very useful tool in the tool belt.

So very excited about that.

Um Justin Meyer and Zan Anderle, they worked on a Django benchmarking project where we're gonna be able to load test different configurations, which is something that's been on my backlog forever and ever and ever.

Um, but they were actually able to dedicate it the time.

And so, you know, we always how is Django fast enough?

Well, it's fast enough for my blog running 10 requests an hour sort of thing.

Yeah, no problem.

Um, what about if we put it under load?

Or what about if we're running ASGI under load?

What if about if we're running with uh free-threaded Python?

What how does that change the thing?

So to be able to construct these different scenarios, run them with with locus to show their actual performance profile, I think that's gonna do a lot more for or to to sort of um to enable us to have a more sophisticated discussion about performance than oh, look how many requests per second it does in this, you know, hypothetical.

Um yeah.

um also

I'm just reading your notes here because I want to make sure we highlight everything.

Uh was Simon also mentoring Annabelle on something?

I don't know.

I only met her at the event and I I didn't uh internalize her surname, but she was a Jjangonaut um alumni.

She went through Janganaut space and um she's interested in the ORM and I was sat across from the table while Simon and Annabelle were working together, and Simon was they were, it was just amazing to watch.

It was the best kind of mentoring, the sort of thing where um we've we've got a essentially a new contributor.

She's not a new contributor, she's you know been through the Janganauts Hospital, but she's still a an entry-level contributor.

She put her hand up at the beginning when I said who considered themselves new.

So, okay, she's a new contributor, and Simon was able to sit with her and they were able to go through things, and she's left the event as an established ORM contributor.

Now, that is that is just gold dust.

Um, and I've got um Jacob Walls, I've got him committed to um auditing our three free threading uh readiness in preparation for 6.2.

Um who else?

Juliana was working on Django tasks, Lillian was helping work on the benchmarking project for a bit, but she was also then working on other stuff.

I don't know what exactly, because she was the other side of the room.

There was people working to update um the uh unit testing for anyway.

I'm listing the the work that gone on, but it's it's it's prolific.

There are half a dozen um PRs on Django which haven't been closed.

There have been some that have been closed.

There's um two or three depths being written, there's you know, progress on long.

Yeah, it's it's literally amazing.

Um, so anyway, that's like a 10-minute monologue about Django on the Met, but I cannot express enough how effective these sprint events are for pushing forward the project.

Yeah.

Yes.

Brains come off of Dangley's software.

I mean, I think the difficulty that we have as a community is making the the marketing case to a company that it's valuable to either either sponsor the event or to send their teams.

And so one other thing that we're doing that's based off um advice from uh Daniele uh Profida is to um uh keep a record of the event, of when what started, what's finished, when's it completed.

And so for the next, for the yet marketing for the next one, we want to say, look, we've done two now, and here are the concrete things that were achieved.

And so there's a real case that you should be sending your team, um, you should be sponsoring the event.

Because you if it's 250, that's different than 25.

I so I think about 2025 is about the right number for the dynamic of the event, which is why we need to have more than just Django on the Med.

Because if we only ever have Django on the Med, then there's too much demand for that event to be the place.

Um Django on the Democrats.

I think if

Like it's really important that we try and help grow this um globally.

We're a global community.

We can't be insisting that you have to come all the way to as nice as coming to the Mediterranean is, we can't be insisting that you come all this way to join in.

That's yeah, yeah.

No, exactly.

And then you were showing me uh Pescara.

So I felt ever more I even got the mug.

Oh, you got a mug?

Oh yeah, yeah.

I love Pescara.

I Okay.

So Abruzzo is like the region.

Anyway.

All right, so thank you.

Um Okay, can I switch gears or is there anything else on the side?

Yeah, yeah, yeah.

I I was trying to follow it online.

I think and I think you know, Danielle's right, you do need to document it a little bit so people can see.

Okay, well they clearly had a good time, but what actually came out of it?

Um so one one canonical place to quickly see just the things you mentioned helps helps companies justify it, helps people you know, I don't think people need to justify it, but companies to see, oh, the yeah.

Like to have um Sam and Charlie from Kraken there.

I mean Kraken are one of the the most serious um users of Django at scale.

And sponsors.

But if they've had to come, they they were sent by Kraken, but if they were had had to come on their time, it would be less well, can I take the holiday and the personal expense and do it?

If the company, even if the company's not sponsoring the event itself, if they can fund the team going because it's an investment in the framework they built their business upon, well then that's a worthwhile result.

Um so this is not gonna be an hour-long episode.

I think the main thing was Django and the Med.

There are some projects, I guess I'll just highlight.

So this is the 29th.

Um, tomorrow the 30th, uh, Air, which is the new uh AI agents in the IDE, comes out uh for JetBrains for PyCharm.

So that's that's been really interesting.

Um been playing with it for months.

Um so not really a third-party package, but something that's taken up a lot of my time.

And uh I'll have more to say in an interesting, non-promotional way, but it's you know, forced me to I think like a lot of developers, I'd been sort of separate panes.

So I have my terminal, I have my git client.

I've been using fork more and more, Carlton, slowly morphing into you IDE of choice.

Um but having it in one interface is the goal.

And it's just a question of how customizable is it?

What does that workflow look like?

Can you spin multiple agents going on at the same time because of the weighting?

So all of all of which is to say like that's been really interesting for me because I've been a little bit, I don't know about behind, but like I'm busy trying to like get stuff done.

I can't endlessly fiddle with the next framework.

Or not the next framework, but you know, the next tooling, because the tooling changes so fast.

So anyway, that's been fun.

And over over the long run, you're like, oh no, I missed that tool entirely, but I saved, you know, numerous hours in in not playing it.

I think interesting see JetBrains take care.

They're clearly amongst the most experienced in the industry about what developer tooling is and could look like.

And, and you know,

the the theory of the thing.

So um, you know, Windsurf.

Anyway, so it's, you know, even if people aren't writing as much, you're still hopefully reading it.

Um so yeah, it's been interesting.

More to more to say on that.

Oh, I was gonna uh one thing that's um we also are gonna be launching uh an AI off mode, which finally, um, so Zed has had this, but we've, you know, many people just turn it all off.

And we weren't able to have that.

We will have that in the new release that comes out um in like a month.

Um, because we heard from enough, you know, educators in particular who say, you know, we use PyCharm in the schools and they get the free trial of AI.

And once the kids get a taste of just tabbing, they're not gonna type, right?

And that also extends to sometimes for my own work.

Sometimes I'm like, I just I don't want the AI, you know.

So you'll be able to toggle that on and off.

And so it's a technically simple thing, but it's important that JetBrains is just trying to be open to like whatever the developer, professional developer wants.

And that's certainly one of the things we've heard, right?

Sometimes I just want to focus on the code.

Um so I'm kind of I'm low-key excited for that one.

And if you if you can't remove yourself from you know something from the candy machine, so to speak, that's just gonna give you answers, regardless of whether they're right or wrong, you can't you can't get the mental space to think them through.

So it's kind of essential that you are able to step away from that and work in quiet work quietly, and then you know, come back and then we'll we'll discuss the the mechanics of editing, you know, as a separate thing.

But the hu human understanding still requires time and effort and a quiet space in which to do it.

So I think that's quite a nice feature or idea.

If you use it in the right way, it's unbelievable, right?

Like I can have it look at every change I've made, all my past versions, the Django docs, you know, and it'll spot like, oh, you used a you know, single quote here, double quotes there, or something changed, you know, for that kind of stuff for for search, basically, it's phenomenal.

The problem is, you know, you sort of like lean back and then you lean back and then you've kind of fallen over because it it sort of lulls you into like you're like, oh, that's but it's it's not thinking, right?

Like I still constantly am trying.

I'm like, just one shot, like a new chapter on you know, tasks in my voice, you know, give it everything it can have in my voice.

And it's still thankfully, it's nothing I would pass as my own, right?

So again, that's that's the sort of nuance in the debate, is like of it's phenomenal for a lot of things.

But if you um yeah, anyways, that's a separate discussion though.

I think for many programmers though, it's it's it's hard because it is so good.

And you know, not everything we're doing requires deep thought.

But if you don't think about it, like, you know, my point is if you can't do it manually, you shouldn't be letting AI do it for you.

Um, but how do you find the space to learn anything manually?

You know, you have to put in some of the reps and some of the space mentally to think about it.

Um anyways, I'll stop there.

But yeah, okay.

You don't need to get it all out in one.

So your project

My my project, I'd just say look, come on, look at some of the ones from um that come out in Django.

So do go and check out uh Django BGT, um Justin Mayer's Django benchmark project.

I think okay, there's a couple there that um have carried out.

Um books, the most important part of the show.

You must have read something.

What have you read?

Um but yeah, I like it a lot.

Um this grassy hermitage, hardly any more than five feet square.

I would gladly quit, but for the rain.

Um well that sounds like fun, maybe not my cup of tea, but I can see how it's like a palette cleanser from code and everything else to just sort of enjoy the wording.

Yes.

Um I have been reading The Wall by Marlene Hossefer, Austrian author.

Uh like f so that so she so is a woman who goes to the countryside and then a this mysterious wall comes up and she's isolated from everyone.

And so through that is a way to examine uh society and herself.

And it sounds like it could be very boring.

In fact, it's really, really, really interesting.

Um, they've made film versions of it.

It's I think it's quite famous in its own little genre, but I've really am liking it.

Um and it's definitely a female.

It I would hard I not to I don't want to be sexist, but it feels like this would be difficult for a man to write.

There's a lot of things that the the author who's a female talks about that seem more female than male in terms of responsibilities and caretaking and all these things.

And so she finds a a cow and a cat, and yeah.

I'm not doing it justice.

I enjoy it.

Like they're fine, you know, but there's such better stuff.

Um but it's

it's very philosophical.

It's sort of like this.

Is there

Like, because all my kids have eaten the Hunger Games books and they, you know, through them and they they really get a lot from them.

But No,

no, my kids too.

So they they couldn't read it.

They're not ready.

I think you you're you're blinded by plot when you're younger, and then when you get to our age, like there's only so many plots.

Like you kind of seen them all.

So it's more the like, what do you have to say about anything to me?

Or you know, combine.

Well the yeah.

No, that's that's the two plots, right?

Hero comes the town, or someone goes on a journey.

That's every story ever.

Uh anyway, so I'm enjoying it.

Um, it's also nice to read.

It's a really good translation.

I mean, I can't read German, but um, it translates very well, very smoothly into English.

So Okay.

And it's short, it's 240 pages.

So you can you can feel like you accomplished something.

It's not these, you know, thousand-page book stops.

Yeah.

So we we are gonna start up the season, so we'll try to get back to our usual cadence.

We'll have some great guests.

You won't have to listen to just us blabber on.

We have guests with things to say.

Um I think that's

That's right.

Let's call it a day.

All right.

Well, everyone, thank you.

If you made it to the end, thank you for hanging with us.

Do check out the links, and we'll have guests coming soon.

So we'll see y'all next time.

See you next time.

Bye bye.