Rust Programming Language Bay Area Rust, 12 Sep 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Rust Bay Area Meetup - September 12, 2019: Contributing in Rust and UI in Rust

Description

- Jane Lusby talking about contributing to Rust
- Adam Perry talking about UI in Rust, and the Moxie framework

A

Alright, alright.

B

um Thanks for coming welcome to the September Bay Area rust, meetup Thank You Mozilla for giving us the space the food this time spread by provided by Google thanks, google.

B

Today we today, we have two talks. First, is by Jane on her experiences, navigating the rust open-source community. This is a kind of a preview of the talk she's going to be giving at Colorado Gold rust next week, Colorado gold rush next week tickets are still open. If you want to go, and then we have Adam Harry, who is kind of also doing a conference talk except this is a version of a talk that he gave at Ruskin last a couple weeks ago at this point anyway. Thank you all welcome, Jane.

C

Okay, so welcome everyone. Thank you. The talk today is navigating the rest open source community. My name is Jane. You already mentioned my tag. Online is y'all, see I will be your guide today, so quick, disclaimer, Manish hadn´t when he invited me for this talk. He did not know exactly what was gonna be in it and it shaped out to have a lot of nice things to say about him.

C

So pure coincidence, nothing to see here, moving on okay. So what's this talk all about there's a couple of goals in this talk. First I want to help new contributors who are looking to get involved in the open source community figure out how to navigate and find issues to work on and get involved and get connected.

C

Second, I want to help establish to maintain errs, who are trying to look for people to contribute to their project, know how to structure it in such a way that it's easier to get into there's not as much friction and how to find people and like what works for getting people involved or what worked for me at least what I'm not gonna do in this talk is talk about how to learn.

C

Russ, there's lots of great people who are doing that already I'm gonna, try and help you find them not be that person today, but hope this talk is just gonna, be about the community side of rusts, not the technical side. So a quick about me, my name is Jean, let's be again and again I'm a C++ developer at scale computing. It's a clustered virtualization company and I work on distributed.

C

Storage I have been working on C++ professionally for 6 years now, I was introduced to rest in early 2018 at one of the hacker Knights and fell in love. I quickly got to the point where I wanted to contribute to the open-source community. I'd actually want to do that. For a long time in general, but rust actually got me there, whereas no one else had really and I'm gonna talk about. Basically why that happened. The first thing was the code of conduct.

C

Now, when I was looking into the rust language, the code of conduct was all over the place, and let me just give you a little more background about myself. This.

D

Is not obvious.

C

I'm, a woman I am also gay and transgender. So when I come into a community space, I worry about things like misogyny homophobia and transphobia, so see coming into the rest community and seeing that they explicitly want to protect people of vary, genders identities, race, ethnicity, disability status and other similar attributes was like a weight off my shoulders. I.

C

What it really gave me was a sense of safety. I felt like I could engage and not worry about. Someone harassing me or if someone did harass me, it felt like the community would have my back.

C

Luckily, in the two years that I've almost two years that have been involved, I haven't had any bad experiences and I think that's also largely to think to are largely due to the code of conduct, acts like preventive medicine and prevents the type of people who who it's designed to prevent or protect against from ever getting involved in the first place.

C

So one takeaway here is mentioning the code of conduct and that you, you explicitly abide by it in your projects, can have a really. It can be all the difference between someone being comfortable to engage and it's not being comfortable to engage so that got me excited about rest, but that wasn't really what got me involved in rust. What really got me involved in rust was this week in rust. So now, if you're not already familiar with it, the rust community maintains a weekly newsletter that contains things like blog posts.

C

The creator of the week call for participation, which is just a list of issues that you can take on, merges change, score of coming meetups and job postings, and sometimes a funny quote at the end, I came for the blog posts because, like I was obsessed at the time, I wanted to absorb any information.

C

I could find about rest, but really had the biggest impact was the call to call for participation, section I have a TD, and so I like when I've been wanting to go up the source work for a while, but my brain is always like later later. It's this big scary thing getting involved is work and work is hard, but here's this weekend rust just every week waving a Friday plug a friendship being like hey here's, something you can work on. Just do it, and eventually one of the issues looked approachable enough and I.

C

Didn't I got my first contribution into rusty. It was just cleaning up some test files that were getting polluted into the codebase, but really get being able to get involved and learn. The process made it a lot less scary, one more thing about this weekend rust is: it is dependent on the community to function. So it used to be the case that the maintainer zs-- to mainly manage it, would go and pick issues that looked interesting and put them in this weekend rust.

C

But they don't do that anymore because they don't want to be unfairly biased towards any one project. So nowadays they rely entirely on people coming with issues that they want other people to work on in order to get them in that list.

C

And if you look at the old posts in the new post, there's a lot fewer issues nowadays, and that makes me personally pretty sad, and so what I want is I want people to do a better job, leveraging this weekend rust and submitting their idea their project things that they have to work on to it. And if you don't already subscribe to it, you absolutely should so.

C

I do I got past the point where I was scared and now I was ready to dive in and try and find issues to work on and I ran into a couple of problems. There's some friction. The first issue was the issue. Boards were pretty difficult to navigate at times. I would go and filter by good first issue and I would see issue after issue.

C

I would go into issue after issue and see that someone else had already claimed it, or someone had mentioned that they claimed it and hadn't done anything firm for a while and eventually I just settled on only looking at issues that had no comments in them, which is it the greatest system, because it's pretty common for people like open issue and then one of the developers to leave a mentorship post explaining how it works so I didn't even get any of those. The second issue was sometimes a lack of feedback or slow response times.

C

I remember distinctly one time, leaving a comment on issue like very shyly being like hey. Would it be okay? If I can take this issue and waiting two days for response and kind of like overthinking, it freaking out a bit and eventually I just deleted the common and tried to find somewhere else, because, like I was too stressed about it and I? Don't want this to come off like I'm complaining about open-source maintainer is not being responsive because they're doing community service, they have no responsibility.

C

You shouldn't expect to work out of them, but that doesn't change the fact that it kind of sucks when you're trying to get involved and you're not hearing anyone.

C

Talk back to you and I I have some ideas for how we can make it a bit easier and if you want to talk about it, come after the talk, I'm not gonna, get into it right now, but I ended up settling on a couple of repos, specifically Clippy rust fixing cargo, where the maintainer 'z were responsive and friendly, and amazing I just want to go special that shout out to Pascal Manish oli, Phil, Kenney, Philip and Eric for putting up with me and helping me subdue my code reviews and all of that and making it a pleasure to work with.

C

So this one is probably not as much of a surprise for you guys, but meetups were a big breakthrough for me, I didn't know they existed and actually, when I was researching. This talk I realized that this specific meetup group is not yet doing a good job about posting us, our meetups in the this week in rust newsletter, which is why I, despite having a meet-up section, I, didn't know about it, but I'm gonna, try and help fix that.

C

But I heard about the meetups from my friend and signed up and started going to all of them and for a while. Not a lot really happened. I was going to the meetups I was enjoying the talks enjoying the food. I was sometimes being social and really shy. So it's like kind of hard to go out of my way to engage, but I was enjoying it and I kept coming, but it wasn't until the meetup at Google, where I really had a breakthrough, and the thing that happened is, it seems it seems trivial.

C

But minutes just came up to me and said: hey, you're, Jane, right and I was shocked. I was like how does Manish know who I am like I felt at the time I felt like I hadn't done anything worthy of notice, and yet here was Manish. Asking me for me, like one of the core team members of rests, the person who runs the meetups was asking for me by name, and so I was like.

C

Yes, yes, I'm Jane and you're Manish right he's like yeah, and he just wanted to thank me for my contributions had done to cliffy and say if you wanted to get involved, some more I'd be happy to mentor you through more issues, and let me tell you like that simple reach um has probably had a more profound positive effect on my life in last year than anything else. In the last two years, I.

C

Probably wouldn't be here today, if it weren't for Manish and his just friendliness, it wasn't the only thing that happened at that meet up I. Also after the talk was by eliza Weissman on tracing and the it ended up being extremely relevant to the system's I.

C

Work on and I brought that talk back and basically gave a simplified version of it for my work and now we're adding a bunch of those features to our project and also after the talk I was sitting there kind of, as always shyly vaguely hoping someone would talk to me and Adam saw me and was like hey. You look like you want someone who talked to you.

C

Why don't you come and sit with us, and so he introduced me to Izzy and rain, and that was like that was the kind of for me when I first got like rust friends, and that was the moment when I really felt like I.

C

Feel like the crying makes it better, but it still.

C

Okay, that was the moment for me when it felt like I'd finally joined the rest community monition. Everyone also telling you that I was already a member. It doesn't really change how it made me feel so before I even left, the Meetup I text I found mini she's username on the get on the discord and I, send him a message: I was like yes, yes, yes, I want you to mentor me like this is I was extremely excited about having someone to help me get more involved.

C

I really want to do the work who's just sometimes hard to like, actually motivate to do it, and so I went home more excited than ever and I had a bit of him kind of I. Think I'm misunderstanding of what a mentor is.

C

What I was what I kind of expected and wanted at the time was basically a manager, someone who would find things for me to work on and check in and help me like stay on topic and get as much work done as I wanted to get done, but- and he did do that, though in retrospect I think that was a lot to ask because, like that is a job, but what he really awkward was much more than that importantly, he was a constant source of encouragement.

C

I would talk to him about things that were, on my mind, like I, talked to him about how excited I was about tracing and how I wanted to get involved in that project. But I was massively intimidated by Eliza at the time and venetie's like no just go talk to her she's friendly it'll be great. You'll have a great time and he was totally right. I I got involved, I've learned more from that project, especially about like high things like unbounded lifetimes and generics and rest then I have from any other project.

C

I had also come and talk to him about, like random features that I wanted and basically or complained him about the bike shed of the moment- and this was he kind of gave me advice on basically like to in this case. It was just to write a blog I like to share your thoughts like. This is a good idea, and this in the specific case it was about a white, syntax and I had opinions, but not specifically about a white I just wanted postfix macros, but he was.

C

He encouraged me to write my to write a little blog post about it. I felt ridiculous at the time, because I didn't write any async code, so I'm like chiming into this massive bike, shed just making it worse, but he encouraged me to do it anyways and I'm glad he did I feel a lot more comfortable logging now and sharing my thoughts and I've been doing it. In addition to that, I was because he's been trying to convince me to give a talk that one needs meetups for a while.

C

He wanted me to do a contribution, tour and trying to talk about all the things I've done and so I've been practicing talks at work, and he gave me the idea to write the talk as a blog post first, which has really helped like it's like every time. I try to do when these talks.

C

It was like this unapproachable task, but writing a blog post for some reason is a lot easier, so I write it out and then I was turned into a slug post or into slides, and it works great speaking of complaining about features that I want Manish fundamentally under changed. My understanding of how to approach open source work prior to this I had been just going out and finding issues to work on and but mostly doing things that other people had said here.

C

You could do this like not necessarily things that I was personally interested in, but whenever I came to him with, like things I wanted in cargo or things I wanted in, like some random library, he was like hey, you should go like we're, gonna, that's a great idea, and it really was a lot more satisfying to work on something that I cared about and something that someone else thought was important, like I got to make feature or implement features that I was gonna use and using them as like way more satisfying than just like writing something than forgetting about it, because you don't really care about it.

C

It ends up feeling like a much more sustainable approach to open source to work on things you personally care about, so I got a bit overzealous at this point. I like had all this energy of this excitement in this approach and I opened a lot of issues and worked on a couple of things, and once I opened three PRS at once on cargo, could them all up?

C

I thought they are great submitted them and Eric came back with the review and it was way more work than I expected that needed to be done, and it was, it was really intimidating, like I, got I kind of got paralyzed for a few weeks and did nothing because I saw like all this stuff that I had suddenly felt like I had committed to, and it took me a moment to realize that I could actually there's not work on some of them for a while and come back to it later and I broke up asks into little tasks and I learned my limit and learned not to bite off more than I could chew.

C

For me, it's one open source issue. In addition to work, another problem or I was having once I got really excited, was I kept on telling people about things I wanted, but I wasn't writing them down very often, and Eliza was the one who actually helped me with this one. Every time I would complain to her Russell Neon, Trees and she's like open an issue and then I would complaint. I would not open an issue sometimes, and then I complained that I forgot something I knew I'd forgot something she was like.

C

We should have over the issue and eventually that sunk in and I started opening issues not just on tracing, but every project that I had an idea for just go to their github repo and open an issue, and it's got the benefit of.

C

Sometimes people do your work for you, which is amazing, but now I have all these issues like around and I was kind of losing track of them, so I needed to find a way to organize them and I actually got the inspiration for this one from David Tolin a he wrote a blog post and published it via Doc's, RS and I thought that was brilliant and was like hey wait.

C

I can use a github repo to track all the other issues from the other, github repos and so I opened one initially I just used an issue, but it turns out when you put a link to another issue in an issue it mentions it and the other one. So I kept something like having a nil notification. Saying Jane put you and PR ideas, Jane put you in PR ideas, and that was a little embarrassing, so I settled on a project board because it's nice to organize and it doesn't notify anyone.

C

So this next one might be a little controversial, but it Manish and Eliza introduced me to Twitter accidentally. It was the a wait posts. Actually that kind of did it. For me, Manish was like what every when I wrote the post he's like here. You should post on Twitter and I'll retweet it and Twitter I'd. Never really like clicked for me before then I was an avid redditor for 10 years, but like Twitter, just Paul when people didn't make sense, but it was a bad one.

C

I was like I could follow people who talked about rust and that that suddenly is like okay, I'm gonna. Do that I followed everyone, I could and I realized that convey that people communicate on Reddit and Twitter are fundamentally different, like red, it's great for threads, it's great for keeping up on news, but it's not a very personal form of discussion that you only have the little username and sometimes people change it like it's one step away from being anonymous.

C

Twitter. On the other hand, you follow individual people, you make friends, you talk to people and there's the picture. There's multiple names, it's a lot easier to kind of remember someone and like make friends and I have made many more friends in the community from Twitter than rustling, hundreds on Twitter and zero unrest or in reddit. So if you want to meet people, I definitely recommend using Twitter and just following every raw station you can find, and the last thing this one hasn't really fully panned out, because it's literally being encouraged to give a talk.

C

So we'll kind of see how this affects me in the long term and I'll make sure to write a blog post about it afterwards, but so monisha I mentioned earlier. Manish have been trying to get me to do. One he's meet up talks tonight, I was I wanted to, but I didn't really feel like. I was ready, yet I didn't if it's the same thing and like impostor syndrome, I didn't feel like I've done anything important and then Jay came along and they were like. Hey I, see all this stuff you've been doing.

C

We have a I was wondering if you would be interested in submitting a talk. Proposal for Colorado, Gold, rest and I was interested. I was terrified as a hoist but very interested, and so I told them about all the things that I was worrying about how I was afraid of like the process and I didn't want to like I didn't know if I'd be able to get a whole talk done in time and they're like no, you just have to have an outline and that they were like they helped me through.

C

All of my fears and made the whole process seem much more approachable and so I said yes, I'll. Do it and I promptly procrastinated for weeks and thankfully that didn't stop Jay, they came back and we're like hey you doing. Okay, you still wanted to talk. I'm like yes, I do wanna. Do the talk.

C

It's just really hard to actually do this, it's scary and they like kept giving me this healthful calming encouragement and got me to do it when, if it weren't for them, I definitely would have just waited past the the time window when I could submit and then beat myself before it afterwards. So the key takeaways of this talk are, if you have a project mention the COC in it.

C

It'll help people engage leverage this we can arrest either follow it or if you have issues you can post in it, post issues go to meetups and conferences, you're already, obviously doing the first get or beat a mentor. I feel like mentorship, had a profound impact on how involved I got and so I feel really feel like. Everyone should have that opportunity and should help the other people around them grow. In the same way. Open issues know your limits. Keep organized use, Twitter do outreach.

C

I think you get the picture so I'm going to take one minute to do quick, shameless plug Manisha's mentorship inspired me to start an awesome list of people who are interested in mentoring. If you are interested in mentoring or wanna mentor, you should definitely check it out.

C

C

It's it's, the union of them both.

B

C

Serene asked on a scale of awesome to amazing: where does Manish land and I said the Union together combine them both he's the.

D

C

B

You can't really embarrass me like that. Just before I have to come up here, it's just not done, but thank you. Jane yeah, Adam I'm, just good I.

E

Agree, the mini she's pretty great, and now we get to see whether all those years in the high tea before I learned how to code are gonna pay off how's everyone doing tonight.

A

That's great to hear.

E

It's a little warm in here. Isn't it.

B

E

It's always the way of it. Isn't it come on this thing?

E

No, it's always the way of it. I'm sorry, I'm, muttering, because I'm tinkering Maneesh help. What do I do.

E

Yeah, the things wrong says: podium laptop. Okay, all the things are on huh most almost all the things wrong: okay, nope! Oh no, this it's the cable, it's the cable, isn't the cable! Yes, it's okay, I'm, just not gonna touch it, but I walks up for the using the laptop um thanks for bearing with me so a piece of advice for the Manisha's of the world.

E

If you tell someone that you'd like them to give a talk, don't tell them to be lazy and another piece of advice for people who are giving talks like the atoms of the world get really excited about things. You can present at a meet-up, don't bite off more than you can chew. Listen to Jane, so I have been doing a lot of hacking and didn't quite plan. My deceleration, four slides prep. So we've got a couple strike. Throughs and I'll do a little vs code. Navigation, but I've also been informed.

E

Another critical mistake on Manisha's part again, my apologies, but I have a lot of time. So, let's get started so I won't belabor this too much I'm. Not anybody that you should know really rust has been really important for my personal and professional development for actually a lot of reasons that sound really familiar to Jane stories.

E

In fact, I found Twitter as like my rust social home for a while, because the rust off topic, IRC channel, had turned into a flaming pile of politics a few years ago, and Twitter was a great place to find people talking about stuff that I cared about without having to do with that.

E

So I'm stoked about rust, I think it has a significant role to play in the future of computing. Oh that actually reminds me that I'm not saying this as represented my employer, and this is not related to anything that I do for work, although is a lot of fun to do some rust at work and before doing a little bit of rust at work.

E

I worked on react, native developer tools at a startup and got a little nerd sniped, you might say eh, you might say: I saw a lot of potential and I still see a lot of potential in in the direction of Technology I. Think you know they they didn't build it in rust, right and, of course, none of the infrastructure to build it in rust, exists or even remotely does and, and so I think, like you know, there might be a role to play for for rust in in the world of future.

E

You know UI abstractions and tools I'd like to explore that, but also I'm, probably very wrong about a lot of things. I don't actually build. Apps I mean part of the reason why I'm doing this is because I'm really picky and I would like to, and that also means that I need to learn a lot from people who do. You know I, see a bunch of interesting, really technical problems when it comes to to interface, work to interactive applications, I think as people interested in important details, the execution of programs.

E

We often underestimate just the sheer amount of complexity that goes into like the human interfaces before it gets to code. That a lot of us are writing day-to-day and that's a lot of complexity. That requires a lot of learning like I said, and one of the ways that I can learn is, if you show me how I'm wrong with projects come write code, we've got to figure some of this stuff out, and so when I say I want to write apps and rust. What do I mean I mean I think we're talking about rust.

E

We generally have some notion of like I, write, cargo, build target and all the magic happens and I think that you know essentially, if you're actually before before I talked about. You know what that implies, for our efforts right, trying to build reusable things and I just want to point out when I was looking for screenshots of AOL email login. This is founders, I, think it's nice if we can all soak in the beauty of our world for a second.

E

If you can't read this that says today's email INBOX looks significantly different from the AOL inbox of your make sure your email marketing has grown with the times and it's tagged with blockchain technology, business, innovation and thought. Catalog.

E

Okay, so human interfaces right call gooeys you eyes whatever, but like the stuff that people touch the things that we put on screens.

E

If you want to be able to just cargo run, that's actually like a really big engineering, artifact, that's like civilization, at work in all of the existing instances of it we have today. You know for context right, like here's, the Mac OS attempt edit or at least their box diagram of some of the pieces that go into that you don't have to read. Any of that. My point is that there's a lot. Here's like a listing of things that go into the application of our the architecture of an android application.

E

If you use like the more modern jetpack framework, it's pretty cool and you know if you, if you look through the API Doc's, like maybe paths, are related to actually things on the screen, but they all have to understand the way that that happens right there, all of the Android flavor and it's the same for the Mac OS flavor, and if you want to be like aggressively cross-platform, so you can just pick something up.

E

You know and run with it and target a bunch of different lat forms all at once like this is flutters architecture, diagram, I! Think it's a really nice architecture diagram. For one thing, it tells you everything you need to abstract an operating system away to build an application, and it also tells you like each of these boxes. You know the event loop interrupts, that's a create thread set up one. Maybe more Dart I slit that up.

E

Well, we don't have a dart crate, but you know if you had a scripting language attached to your UI, you would probably have a few crates in there gestures, multiple crates, painting 20, who knows these are like complex pieces of software, each of them and they all have to interoperate and I.

E

Think those technical realities of that complexity are part of why we have large frameworks like this typically right and if, of course, there is kind of this exception that, like people, kind of compose their own application framework on the web and the web is really cool. It's like the only thing of its kind in all of history and may forever be.

E

Who knows, but it also because of its very cool properties of backwards compatibility has a difficult time serving everyone's needs all the time it can take a long time for it to achieve good performance for certain kinds of interactions and like if you look at the global population of people using devices they're, often pretty poorly served by the mobile web, so I don't think it can be the only thing that a community bets on if you want to be like building applications for everyone and I, think you know if you look at the size of these engineering projects.

E

Aside from the web, which you can kind of write a library really quickly for I've already done that, and you know it means that something that's like flutter scale isn't coming with flutter polish. That soon, at least by my by my standards, I think that there are multiple, really really really exciting, like in pure rust, native GUI efforts happening and I want to write, boxy bindings for all of them, and that's the name of the project talking about tonight.

E

But I think that, like to target, you know desktop mobile web embedded stuff that people might want to write UI for who knows the kinds of the the breadth of targets that Russ stations tend to want to embrace. It's just really hard. It's just a big task. It's just a lot of big tasks and coordinating them as hard and we're gonna like do that as a community, but I think that basically and bear with me this is aspirational.

E

I think we can kind of digest. This incrementally, like I, think we can push forward on pure rust experiments projects. Pursuits I, think like druid in particular, has shown lots of really exciting stuff happening and orb ck has a bunch of great demos and they have like a whole desktop environment. A zoo looks really cool. The the ice renderer looks really cool.

E

It's like these are all vital and important for our community to do, but we can't only vent if we want to like ship stuff, at least if we want to ship stuff in the near-term, and so this is why I think that kind of exploring a you know, libraries and frameworks approach is interesting and I think that with rusts we actually have some kind of historically unique ingredients for doing this. With respect to like complex apps that you hold in your hand right, we have a community that really values deep research.

E

I didn't put this on the slide, but you know like people who think a lot about problems and experiment a lot, and you know there's there's a place for that in our in our in our circles and we distribute those artifacts that they create and we have a type system that makes it really easy to build abstractions on top of them and to do so with low overhead and I.

E

Think that you know, if you look at the critical performance properties of lots of you eyes, you know the scaling factors that make you know frameworks necessary. Might we might be able to kind of relax those a little bit for us might be able to shard the work a little more, and so my aspiration is that if we start with, you know bindings to big projects.

E

While we have a bunch of these exciting efforts going on that, you know we can kind of build like rust applications and also pave the way for the future and I think that, concretely, what that looks like in practice is a bunch of code that you can use no matter the context right or at least across several contexts, which you care about right and so I've been attempt making a bunch of attempts at this for a while. This is the one that seems worth sharing, calling it moxie.

E

Originally I wrote a JSX style macro that I called mocks because it made a mockery of XML, and this seems like fun, project name, and so the buzzword soup that you see on the right is like my attempt at distilling what I think one really interesting piece of critical infrastructure for rusts community UI projects might be, which is like a UI runtime.

E

If you're familiar with that's likely the kind of niche in the ecosystem that they intend to occupy, I, think that, like the rust answer to this part of an application, is probably you know, lower level a little bit less opinionated, etc. So this is kind of my attempt at playing with that and lets.

E

You know maybe dive in a little bit to some details, and the key bit here is that all the demos that I had to show are from this library of Moxie Dom, which is using this run time to manage Dom notes, but excuse me, and if one of the core the core thesis here, is that these semantics and like being able to express the shape of application and rust could quite easily port to other to other platforms and like I, have some.

E

You know some half-baked experiments with Windows and with Pathfinder and stuff like that right and so the the open road lies ahead.

E

See if I can remember what slide this is: oh yeah, okay, so this is where I spent the last couple of weeks. Hacking, instead of showing instead of you, know, polishing my slides and and also a quick warning to anybody watching the video this will likely possibly be live before the video of the rest confluence we're doing kind of a fun time. Traveling thing the code that I'm not showing here is, you know, going to like be at a later publication and anyways. So what are we showing? First, we're gonna show the counter. First.

E

Very professional, okay, so we, but we do some like basic prelude. This is all wasum by the way and rusts wasum toolchains pretty sweet.

E

So if we let's scroll out of this, is this visible, okay, cool in back awesome? um So you know basically, if your he views like react hooks. This is probably the closest analogy to like the style here and these macros are.

E

You know, they're aware of their call site. Basically, so if you had a if you had a second one here, these would be distinct state variables that allocates you allocate some some persistent state for your for your component here and we format some some text for it, and then we create a button right. This button has an event handler and on on the event handler being fired. It will update this state variable. This counts right. So when.

A

E

State will be incremented yeah and then, of course, because you know we have to test lists and stuff and like you want to have a couple of little things that will break for you if you do something dumb and move that, so this is mock. Crs, that's bigger to this is published from top of tree from CI. There's a discord you can join. Obviously, I hope that this can be a community space, but it's very early. Oh there's a code of conduct. I will enforce the out of it.

E

Thank you for the reminder, Jane, okay. So here's a counter right so boom one. How high can we count? I, don't know this is the pinnacle of modern UI development. That's that's good!

E

So yeah, that's the counter right! I! Have many schemes to you know make this a nicer to write I, think one of them is if there's there's already been an IRC approved for a lingua feature. That I think will make it possible to eliminate all the macros, so you can just apply an annotation to any function and then it becomes topologically aware, which that's a phrase that I hope catches on to explain the weird stuff that I do and I will explain that in a minute so yeah. This is a that's.

E

So that's kind of the minimal example there right- and this is you know in this. This scheduler in behind this function calls a few hundred lines of code and it you know it respond. It only runs code in in response to actual state mutations. So not all events will cause the runtime to re-execute and it does so like with fancy requestanimationframe based scheduling. So you shouldn't be blocking the main thread most the time it's integrated with the tracing crate. So soon I should be able to have like a tracing timing.

E

Thing that'll tell you, you know if you had like. If you blew your requestanimationframe budget, we can give you a breakdown of which parts of your component topology were causing that problem. Things like that. So you know it's very early days, but I think that there's some you know some promising stuff happening under the hood and let's go back to some slides, yeah. Here's a video of the same thing. Do we want to watch it again? No, okay, so decomposing the buzzword, soup a little bit right, a little code sample little demo.

E

Platform, agnostic and lightweight: what does that mean? Well, it's rust! So if you don't put platform in it, it just is right. This.

A

Is a simplistic.

E

Explanation but you know basically the idea is that if you can define a bunch of the key semantics to share between all the libraries in the lower level, crates that don't know anything about it, DOM or a tree or whatever, then you can reuse those in interesting places and since compilation times have been a subject of conversation recently, I would like to keep them fast, etc.

E

But you know oh and also I, really really really want to make this work without like a threading model, or you know, if you just have an allocator things like that and I almost made it happen. But then it was gonna be a whole yak shave and so that's for a future. A future revision but, like I, think it's very plausible that you could have some. You know kind of mid-range, embedded thing that actually has some graphics to do that.

E

You could still use this this one time in I haven't benchmark to sufficiently to prove that out, but, like it's plausible, so the other half of the catchphrase declared a Qi runtime is not a phrase. I think that anyone ever used until react like took over everyone's brains.

E

You know the the the piece at which it's become it or it the concepts that its popularized have become kind of the de facto way to describe new UI projects.

E

I think is telling like I said: I, don't actually write very many applications, I've done some web development and things like that, but the the social proof seems very strong to me and also there's some really nice technical problems, technical properties like not having to think about time when you write the description of your UI like it, it's just right now so before we get into like what that means in terms of an implementation or what we do or things like that, let's just talk about like the role of UI runtime right: it's not like the application logic.

E

It's responsible for oh I'm, not even presenting yeah there we go cool, so it's resolved responsible for presenting a bunch of visual items right. These are Dom nodes on the web. It's, like you, know a little sub canvas and some UI frameworks right. A little rectangle doesn't matter. It's tree things right they nest and then the the important person, the VIP, the human, does something, or maybe something that they've asked to act on. Their behalf does something like an I/o driver or network response, and that requires updates to the interface right.

E

So you have to change something on the screen and this needs to be a magical lie of a continuous, constant object like it should be a persistent thing. But computers, don't really do persistent things. They do things in loops right. They do the same thing in loop and for you why we usually talk about these being scheduled as frames.

E

So that's kind of the 10,000 foot view of what we're talking about. This is like a very bad diagram of the like UI pipeline in like a traditional Model, View controller type thing and almost all of the applications up until like the react phenomenon, just skyrocketed were written in a style where you would manually create this tree, and you would write code that updated the state of the tree based on events right, and so the UI description was kind of a property of, like all of the accumulated mutations.

E

You've made up until that point, there's very good reasons to write UI this way. For one thing, it's a format that the computer understands like you can map it to the GPU, also like there's a lot of weird sophisticated stuff that you end up having to do if you want to describe it differently and so to keep it here is that you have something you know that you're manually, bringing through to the next frame right and the world has decided they would like to describe the state of the UI right now.

E

You know, like this moment in GUI, was really popular like six seven years ago among game developers. This is, you know, very pleasant code compared to some big QT tree to manage, although qml is pretty cool like everybody's doing the thing right- and this has some limitations because you have to like write directly to GPU, essentially etc, and some really smart people came along and they said, I don't know if this is actually the history, but they said we would like to do this on the web.

E

We have these Dom nodes, they require explicit mutation, but I would like to just kind of write HTML and have my application. Do it for me right, and so they have this state that they manage and they bring it along for you, and then they do this reconciliation in the runtime right. The runtime goes through and says, like ah yes, I see that you would like to have this div appear here and it was not there before.

E

I will create it for you and you get very cool things like you can essentially interleave what pretends to be HTML and your JavaScript, and it works like that right and so I think. If we want to go beyond that or if we want to say like where does rust have a role to play in this kind of paradigm, I think it's. Basically, we just have to admit that rust is harder to write than JavaScript and then basically every other language that people actually want to write.

E

Applications in right, like is a space we're serving row stations. Interests is an awesome achievement and I will be happy just to do that, but the potential that I see in our community and in our technology to go further than that to kind of you know take over the world comes from just admitting, like we've got to make this a little bit easier to match the domain I think. But that also means we have an opportunity to find unique ways to invest RIT's rusts properties like so.

E

If you have to pay attention to allocations right, like they show up in your syntax, that's not the case for lots of languages, especially ones that people like to write nice, sugary uija, then maybe there's a way that we can take those same properties and you know achieve new interesting things or at least achieve productivity, that you would not find in a different environment right and, of course, maximum portability.

E

So the basic idea here is both react and the MVC mutable object graph thing, except we have some tools for making all of our updates to that. To that you know stable tree incremental, and that's you know, broadly speaking, like people talk about it like caching or incremental computation, moxy talks about it in terms of memoization, because it's at specific call sites for specific function. Indications as I showed you in the slide earlier, and so then this this storage of memorize items comes across.

E

And how do we do this? How do we describe you know things just like their HTML or something similar in in line with our Rusco hoti. Here Wow, yes, Bob I will come down. Okay,.

E

So if you squint a function, call graph, it's.

D

Like a tree right.

E

Yeah sure, okay, so let's do that? Okay, we have some functions here, their names, don't matter. Let's pretend that indent is we'll call it a topologically aware function and it knows its depth in the function call tree yeah and it well. You know, like it says, on the tin indent, the string, that's printing according to its depth in the tree, so we have the root. Every time we enter a stack frame, we call a function. We create a node in this tree. This is like not concretely we're not actually doing this.

E

Don't worry, I'm, not totally bonkers, but this is kind of this is conceptually the Scopes we're creating and they have concrete identifiers in the runtime and- and so this is the tree construction.

E

So we allocate a root node, it's not at any depth it's at zero, so it doesn't get indented when it prints and then it calls a I know. This is extremely complex. Algorithmic computation bear with me. We enter a stack frame. We allocate a node in the tree in the abstract tree and aha, we've indented yeah and similarly, we put back out go into P creates a node. We have an indentation enter a loop call.

E

Some other thing has another side effect, and you know if we, if we extend this, this concept right control flow should just work and I mean I, know slide, so we're not doing anything impressive. But- and you know this is this- is the thing we would want. If we were just going to use function, calls to describe a tree. We just want the control flow to look like the tree as a tree to look like the control flow.

E

It depends on which direction you're squinting frame, and so you know we have C sub two right and you can see kind of we we've accumulated. This might be an abstract tree over here, but you can see that on the left in our our fake console, we've accumulated side effects according to this function. Call that map to the structure of both the syntax and the execution right, and so this is, you know this mapping from execution of function. Calls into some structure of side-effects is the core bit that we're doing here.

E

I know that that's really out there. Let's talk we'll talk more so this tree or this topology is implemented in this crate, underneath Moxie called toe, oh very originally named, and, and so the idea is kind of like it's. It's making these stack frames and their position over time, relative to other stack frames right.

E

So it's like if you have a stack, you're, pushing on and then you're kind of move over in time and then another one on we're a if I this notion in a concrete identifier right, so you can call the function or inside your function at any time and say where am I and that will be unique to your position in this in this tree? The way we do this is we essentially trace the path of these executions down right. So when we call a function, it has an identifier.

E

If there's no parents for it, like there's no topologically aware parent gets a 0 by default, so you can see the 0 up there by the root and then each time you enter a function that call site for that function has unique identifier associated with it right. So your your function has excuse me its source location, but then, where it's invoked has a unique source location as well, and so the ID those properties all hash together.

E

Basically right and the important bit here is that if we do this right, we build this tree walking these these functions. We can do this in a loop, hey, there's a UI runtime, and if we do that in a loop, those IDs will be stable across executions. So you know we call the loop once we create the tree, we call the loop again if the trees changed. Maybe we should do something if the tree has n't that's a lot of work we can skip, and so basically the moxy crate takes this way of generating identifier.

E

X', and you know it's pause waits for events when an event occurs. It calls the root closure. You provide it and then it goes to sleep again, and so, when it's calling this root, we have some tools that create kind of a runtime environment conducive to describing UI.

E

Small sight, if you know react, I mean it's fairly popular, so I figure some people will here well. This like topological, aware functions, Maps very closely to react hooks except you can use them in conditionals and loops and recursion, etc, and Moxie kind of maps on to react implemented with hooks and context. Basically, because topo also has an environment that it can create, which is like scoped global variables, and so you know we we have these revisions or frames around ten gets invoked on each one.

E

If we just naively create side effects, the runtime is very inefficient and much much worse than anything. You would actually build an application with today. So let's not do that. We have a few primitives for doing so. You can say: I would like this to happen once for this scope in my tree. I would like for this to happen anytime. This variable changes from the previous invocation right, that's the the memoization, and then you also because these values are stored. In kind of you know, outside of your your code, just ready there waiting for you.

E

We also have a single place at which we can garbage collect them if they weren't referenced in the last revision and in rust, you can implement drop on types so that, when they're destroyed, you can run arbitrary logic and in this way, I think you know, there's like some interesting paths to essentially moving into the language things that are normally UI framework concepts that you get implemented per framework.

E

You can build. You know persistent state variables in the memoization store that you know take concrete updates and you can use this to build a nice abstractions for the Dom so that you know things only rerun when the state changes, etc and so I've built that to do MVC sample with this I know it's very hot.

E

To do MVC and behold, so we can check things completed. We can delete tasks, you can add new tasks. This is like a year of work, so pretty stoked about it.

E

Thank you. Thank you. There are some. There are some for visas, which is that, like this, doesn't actually do the memoization yet so, like the fact that this all works and isn't totally broken is because it's.

C

E

Yet, there's just that couple of bugs that I'm still still working out and but all the pieces are there and memoization the memoization storage and call site identifications that are necessary to construct this. It's just they're not being used to maintain the Dom tree across revisions, yet yeah, so you know you can know, did I break it. No I didn't break it. There. We go switchin active and completed its ah okay. So that's all very boring.

E

This is actually live online though, and it's running webassembly in your browser and there's no JavaScript except I, think like a little shim, trampoline thing yeah. So here's this we have reload. No, that's that's for your development! That's not running! Oh yeah!.

B

Here's the JavaScript.

E

Yeah so there's I lied, sorry, there's some JavaScript, it's the Waze impact, javascript plus and wait so everything else, the rest, that's pretty cool and some CSS thermal stuff, and so this is. This- is the tree made by like this weird branch, where the syntax that I've been showing is coming from, and you know we have so some other things along with project. Just point out. This is like a server. That's running some cargo watch tasks.

E

The demo curse had happy somewhere happen somewhere, and so you know we like. We have some cargo watch based things and when files change that server will you know reload our tab so like if I were to break something?

E

This might work.

E

Here we go so this is the root of the application. Let's change the default to do that. It loads with.

E

One thing about this workflow is that I have too many tasks stacked up in the Watchers, we're compiling to do them. You see now cool so does that reload for us? No it's! It knows why, probably because file system notifications, don't work very well when you start The Watcher, before suspending your laptop.

E

Okay, well, the demo broke, that's expected!

E

So let's look at how this is implemented a little bit yeah.

E

Here's you know a little bit more complex state than a counter if you've done JavaScript stuff. What we're doing here might be familiar is like a Redux atom right so like this is the application we have the state for the application, which is you know, the visibility and the list of siddhu's that we're gonna mutate right and then we provide these. As kind of you know, Singleton's to our environment, we create a div and we call some component functions or some.

E

You know some topologically aware functions the header in main section there, and if we, you know, look in like when we go and render the list of of things we have a bunch of junk junk, junk, drunk drunk drunk no in here we have for loop, nice, and so that's how to do items are rendered. As you know, we loop over a filtered iterator and I.

E

If this isn't, if you haven't done collections or filtering or things like that in, like reactive UI systems, this seems like very obvious and boring, and if you've been in JSX land for years really I'm telling you trust me, and so this is kind of the basic structure of things right. There's some other, possibly interesting stuff like in handling inputs can be you little weird, that's not so bad.

E

You know we create placeholder and like have some Dom interactions. This is all using web cysts by the way, which is fantastic and my favorite feature rust doc. Links to mdn for everything I would ever want to know about the web. It's like the fastest way I can find web Docs.

E

You know so like if we're having having an input here, we have a couple of different events that we have to handle, and so you know these that the Moxie Dom setup right now the event handling in order to kind of guide you into doing less work only handles events on steep pointers. Essentially right, so you allocate a state variable, and you say here is a pointer to it. Please mutate! It when the Dom has an it receives an event, and so they, you know you get some event and apply the other.

E

You guys you get provide a new value for this, for the text input field right and similarly like on a key press. We have some handling logic. You know it's all pretty. It's all fairly boring, there's a little bit of steep mismatching stuff with the Dom here that I need to sort out, but I don't think it's nothing too bad, and so this is like code that works and does stuff and it's fun. But it's like not very much here's another! Oh, that was what I showed you well done. Good, remembering Adam! Here's video!

E

Yes me, yeah I, didn't do a very good demo. So, let's you know we'll watch this. So if, if the internet agrees yeah so that you know this is like very, very boring, I did demo Moxie or Oh scoffs. That's good! Didn't wanna! Forget that yeah! Okay, that's enough! So where are we today? Why am I talking it's not working? It's very broken. It's very early. Well, I would like to take Jane's advice. Now that I've heard it all in one list and I was kind of heading in that direction.

E

There needs to be more people working on UI, stuff and rust if we want to build applications. That's just a fact: I am one person who has a day, job I, actually kind of like the stuff that I'm working on right now. So like I'm, not you know charging full into this like full time. I'd like to have evenings occasionally so I've got to find other people.

E

I have to do the thing that got me into this community in the first place, the thing that like helped me grow as professional and like brought lots of new perspectives to me- and you know, okay, so I actually I'm pretty excited to like encourage people to come and contribute.

E

Please come contribute, but with the with the pervy visa that, like it's pretty mature, you know I, think there's some good ideas, or at least interesting worth examining before abandoning you know, performances, unknown, I, think there's a path to making it wicked fast, but I haven't started that at all documentation and testing is very early. Although been doing more on that, since I wrote, these slides originally syntax is improving, but there's no JSX and I would really really like typed HTML to integrate nicely I haven't started hacking on that.

E

Presumably I will pest your Bodil at some point, but I think that's kind of one of the things I find exciting about the rest ecosystem. At least. Is you know, here's here's, this developer experience I have in mind. I would like to take at declare to you. I've run time and I would like to write my syntax for it matching the platform targeting, and this doesn't have to be one monolithic, crate right. This doesn't have to be a couple of things and we don't even need language support like it's all just there.

E

So that's cool, that's a cool thought I'm not going to build cross-platform elements for this until somebody else does probably I think like there's just a lot of work to be done, making it possible to build rust apps with that particular platform in mind, and eventually we will get to a point where it's easy to take the abstractions from those and kind of voice them into libraries and yeah. There's nothing built for it. That's not in this talk, so you know just temper temper expectations. There's like a jillion and a half things to do.

E

I would love your help.

E

I turned the sound off on this because I didn't know. I didn't have a chance to do a volume check, but it's a pretty great song. If you want to hear the crab rave song.

E

Yeah I think that's like what the future you I could look like, potentially just think just think so: yeah Moxie dot RS. Here's the website that QR code goes to the discord. I still get push notifications for every joint and every message it gives me so much dopamine and I would love to share that experience and if you would like I will take questions if we don't all want to run into the evening heat area evening cool from the heat you talking, cooked thanks.

E

E

D

So I've not written much react, but my understanding is that in react, the persistence of state to cross frames is dependent on basically the details of the virtual Dom dipping algorithm or the competitor for the component. Dipping y'all, rather rather the.

E

Sort of like there's there's, you know component state and then there's the runtime State. So.

D

I'm talking two components: data sure.

E

D

Like a choice of whether to reinitialize the component state or use it from the previous one, yes is dependent on the diffing algorithm. Yes right does this have the same property? No, so how is State House component State? If there is such a thing, there's.

E

Not such thing so.

D

Then what was that state background? A state.

E

Variable: okay,.

E

So, okay, so I had a component rate and in fact, when the video from rest cough comes out and embarrasses me thoroughly, I have some syntax in the slides that has like here's, a component rate. Here's how you implement it. My experience implementing just a couple of things quickly turned out to be like yeah.

E

You just want functions, and this is also, incidentally, the direction that a friend of mine who works on jetpack compose a couple of months ago, like nudged me and was like we're doing this like component abstractions, are overhead important for certain things, but not for your thing and I found him to be convincing, but much after he tried to convince me. So thanks Leland.

E

So as a result of that, it's like functions and their variables and some of those variables happen to be memo izybelle. Some of those variables happen to be mutable but out-of-band with the current execution right like by handing it off to an event handler and those are the primitives of the right time. That's it so like. If you I, don't have it I, don't have nested memoization working yet.

E

So, if you want to like skip internal portions of the tree, it only works on the leaf nodes right now, but once I have dependency tracking for those, it should be pretty straightforward and then you'll be able to just like say well. If this variable hasn't changed, don't rerun those functions and then those side effects will be memorized as a result, because we'll have stuck their struts representing that having been done into the metallization store, with the drop handler to undo their effects later right.

E

That's re I agree with a lot there, but that there's no like component trait there's no like hosted state from the runtime. The dipping, in fact, is done entirely in the leaf nodes that mutate, the host environment, like there's, no there's, no, there's no I, guess technically and it's in memo, which is to see if the memo capture variable has changed. That's the diffic.

D

E

Any more confusing answers to give I'm trying hard I'm sorry.

A

Either I'm wondering what the what feedback you got at Ruskin regarding the declarative style in this model like when you look at other frameworks that venture to go into the declarative, UI is including I. Think one of the most recent ones is I. Think it's pronounced by Rick, Harris I believe.

D

A

He decides inspiration also by I, can name a guy named Paul and and the idea that you have like this quote-unquote destiny operator, which takes the role of your state, annotations, so I supposed to say any state extra H mark. And then you say this is the variable that is going to be maintaining. That is supposed to be always shown in the UI.

A

Instead of having that having a specific operator that says whatever you're operating on in this specific variable, it ought to be always displayed up to date to the user and seeing what you've done. I was wondering if people at Rosco mentioned something along those lines saying well. There is the notion of declarative programming, not just an e in the world of UI, but in general, but to deal with that, you have to really rethink the notion of what an assignment operator is.

A

It really changes a lot so I'm wondering if anyone picked up on that or what the type of discussions you had in in that space, where, if any so.

E

I had a brief interaction with Rick about felt a while ago on Twitter, maybe I. Don't remember and I actually think that I think that the model I saw him talking about when I watched. His talk seems pretty similar, actually so that democracy we're like under the hood. It's just poking at mutable, stuff mediated by some nice trigger for it I think yeah I mean an assignment operator for a state variable sounds great. I I suspect that it will be awhile before our approaches, our meaning. The Russ communities.

E

Approaches have matured enough for that to go into the language and, in the meantime, I think it's pretty cool that we have the ability to do this with macros and stations, and things like that, to the extent that we can also I'm kind of interested in exploring this space of like achieving very declarative syntax without breaking mental models.

E

Right, like I'd, like the guts of it, to kind of be bare, not because I think that that's like morally virtuous, but because I think that you know, shifting intuition is just hard and the way people debug is a certain way and so I think that the like, having kind of top-down imperative control flow, where you can reason very very directly about the execution order of things in this description of the UI I.

E

Think there's actually valuable in that value in that, especially when you consider that it's very straightforward to make like lifecycle operations happen with drop handlers right, like I, think you can actually do declare DUI, but you know down-and-dirty imperative, Lee, essentially with rusts and so I think, that's kind of like the Engel that I'm looking at right. Now, that said, yeah I mean I, think most frameworks lust for language support and UI is a difficult domain and I wouldn't like rule that out as something to do in the future. Have I answered your question.

E

Okay, cool I've been informed that we're gonna keep it to a limited number of questions. We cab it here.

B

E

Yeah I'll hang out for a few and also there's you know: discord, QR codes, very 2011, sorry, 2019 yeah, but I'm here confess to me I'm online. Compare me. She email me at I. Just set this up lol at a NP lol.

E

I will schedule coffee with you or chat, and I am a bit time constraint. So I haven't signed up for James awesome mentorship list, but within my little realm, I am of course very available and would love to to offer a direction in her opinion or whatever, if you're interested, but not sure where to start, and thank you very much for listening to me talk to this long.

B

Thanks Adam and we still have some time so feel free to hang around talk. You can ask the speakers any questions you still had and as always, if you have any ideas for talks, let me know you can message me on Meetup. Thank you again for coming.