A great aspect of working with Elixir over the last 10 years. One of the biggest selling points is that it just runs. It doesn't really break in the way that some other languages will. I'm using break, the air quotes are on the word break very lightly. What would you say to somebody who is skeptical that it will be similar to that in the next 10 years?

So based on like Elixir being extensible, I think one of the things that has allowed us to be unbreakable is that if we want to change something in the language we are not comfortable with, we can change the language because it's accessible. Most of the language is written in Elixir itself. So it's very rare that we can paint ourselves into a corner where we're like, well, the only possible answer for this is to break existing code. Throughout our history, there is only one issue in the Easter Tracker for Elixir tool whenever that happens, which is a very silly thing that it's going to affect very, very few people, which I think it's a, and well, let me say what it is. So we have module attributes in the Elixir, right? You can say add something and you can put a value in there. And add doc is a module attribute where you put a documentation. And because in Elixir, everything's an expression, it means that add doc has to return something, right? And what it returns is, I don't remember exactly what it returns, but whatever it returns is not enough tracks as information about documentation. So the proposal is to change the value of that attribute to something else. So it's a very small change, like how many people rely on the value of that doc. And that's the only one. So I think it's a pretty good track record. And whenever, boy, we have a lot of deprecations, right? Like if you go to our change log, or release all the deprecations that we have done throughout the years, it's a long list. But that's exactly the idea. It's like, you know, if something we feel like, oh, this was a mistake, we can deprecate it. We can improve, we can provide new APIs, and we can do all of that without breaking existing code.

So sometimes speaking of deprecations, you might work on an older version of Elixir for a while. Sometimes, you know, upgrades are kind of prioritized, lower than feature development. So you might find yourself kind of dealing with a large batch of deprecation warnings. I'm curious if you have kind of suggestions for folks as they're kind of working through those so that they're not kicking the can down the road for another year, sometimes. Do you recommend working on deprecations in isolation, like a dedicated PR or something like that?

Yeah, that's usually the process that I go through. I update and then I go through the warnings. And most of the time, the warnings, they are very small fixes, right? So I think the next version has probably just one and two warnings. And like the warning is very small. We have the log levels. And then Airlane added like five years ago, they added like syslog support. And syslog is like emergency critical error notice. So all those there are nouns. And we had logger worn, which was a verb. So this is a thing that we haven't alternative for that for a while. And you just have to change from logger warning to warning. So they are typically small changes that you can go and do it with no problem. And usually I reserve a PR for that separate PR or sometimes when it's super small, just do it all together. And in the same path, which has been typically the case with Elixir.

So yeah, that logger.worn, that's like my default. So I've been trying to like train myself to say warning with dializer and language server. I get the squiggles, the compiler warnings that I've used a deprecated function. So that's always helpful to see it in real time as I'm updating my kind of mental process, you know, writing the code. But I think our listeners might kill us if we talked about deprecations for an hour. So going back to ElixirConf and ElixirConf EU and the things you were talking about just a few days ago, looks like your focus has been largely on live book and Elixir tooling. Is that fair to say?

Yeah, I think so. And machine learning,

yeah. And you mentioned it right at the top of the of the talk that was the biggest EU conf or Elixir EU conf ever. Do you have any idea how many people were there?

I believe it was between 550 600.

That is, that's

a large conference.