058: Rust and Going Into Business with Carol Goulding
We talk to Carol Goulding, co-author of The Rust Programming Language Book, about going into business in Rust, Crates.io, and organizing the Rust Belt Rust Conference.
- 00:58 - Going Into Business Using Rust
- 03:42 - Getting Paid to do Open Source
- 05:31 - Prototyping Projects in Rust
- 06:21 - Why Rust? (Benefits)
- 09:58 - The Language Server
- 14:52 - Error Messages
- 19:46 - The Rust Programming Language Book
- 23:35 - Crates.io
- 27:41 - The Backend
- 31:11 - Working with Rust and Ember Together
- 33:31 - Rust Belt Rust Conference
- 35:59 - Integer 32
- The Rust Programming Language Book
- The Frontside Podcast Episode 51: Rust and APIs with Steve Klabnik
- Rust For Rubyists
- Working Effectively with Legacy Code by Michael Feathers
- Python Koans
- Rust - exercism.io
- No Starch
- Rust Belt Rust Conference
STEPHANIE: Hello, everybody. Welcome to The Frontside Podcast. This is Stephanie Riera. I am a developer here at The Frontside and with us, we have some very special guests. We have Chris Freeman who is a former Frontsider. He is a developer at a startup here in town in Austin called OJO. I'm going to let Chris introduce Carol Nichols.
CHRIS: Hi, everyone. Today we've get Carol Nichols. She is involved in a lot of different things related to the Rust programming language. She is on the Rust community team. She is the co-author of the Rust book. She's the co-founder of a Rust consulting company called Integer 32 and she's the maintainer of Crates.io. How are you doing today, Carol?
CAROL: I'm great. Thank you for having me on the show.
CHRIS: Thanks for joining us. I have a lot of questions for you. I'm very interested in Rust but I am especially interested in some of the stuff you're doing that's kind of ancillary to it, namely you decided to go into business using a pretty new programming language that in some ways, I think is a little bit niche-related to some other things that people might go into business for say, web development. I was hoping maybe you could talk about what is Integer 32? What led you to starting this business? What kind of consulting work do you find working in something like Rust?
CAROL: Integer 32 is my husband and I, Jake Goulding and we decided to form this company because we really wanted to get paid to work on Rust. We think Rust is really interesting and that is moving the industry forward and we see a future in Rust. As far as we can tell, we are the first Rust-focus consultancy in the world, which either makes us trendsetters or really stupid. I'm not sure about that yet but we're figuring it out.
We consider ourselves pretty qualified to be running a Rust consultancy. As you mentioned, I'm the co-author on the book. I've been working with Rust for a couple of years now. Jake has the most points on Stack Overflow in the Rust tag. We’ve got a lot of experience in getting to know Rust. We've been watching the development, helping people learn Rust so we are offering a bunch of different services.
One is to build an MVP or a prototype for Rust so that companies can evaluate whether Rust would be a good fit for their problem, without diverting their whole team to be able to learn Rust enough to evaluate it properly. We’ve done some prototypes. We're also interested in doing training and pairing so we have some training, things in the works. We’ve also gotten some jobs that are adding to open source libraries in Rust. The ecosystem is still being built up and there's a lot of libraries that do whatever the person who wrote them need them to do but they're not feature complete so if someone else just needs that extra feature on some library, they can pay us to add it if they'd like.
One of the things I really want to do with my consultancy is have our proprietary work subsidize our open source work because I really wanted to get paid for open source stuff. We have a different rate that we charge for a proprietary versus open source. We’ve had a few gigs that are adding stuff to open source libraries and I love those because we're not only benefiting the company who needs something but we're benefiting the entire community.
CHRIS: When you say you work on an open source thing, do you mean like a company that happens to be a consumer of an open source library would pay you to add a feature? Or is it the maintainers of the libraries themselves are coming to you and hiring help?
CAROL: So far, it has been the former but we have talked to some people about the latter but open source projects typically don't have much funding. I think that's a little rarer but definitely, were open to companies paying us to add what they need to a library.
CHRIS: Has there been any friction there like you kind of showing up and say a company is paying us to try and add features to your project? Do the maintainers ever pushback or are they very happy to just have someone helping?
CAROL: Yes, so far no. All the maintainers we worked with have been amazing. We’re not going to come in and rewrite the whole project. We’re going to come in and work with their style and make any modifications they want to be able to incorporate into their library. But as I said, a lot of libraries are gotten to a certain point and I think the maintainers would like their libraries to become more feature complete but everyone only has so much time and you don't necessarily know what's useful to people but this is a very, very strong signal that this library would be useful to someone if only it had this little extra thing. I think most maintainers are open to making their libraries more feature complete to be more useful to more people.
CHRIS: Yeah, that is a pretty sweet deal from the standpoint of an open source maintainer. It’s nice enough when people show up to help at all. It is especially nice to show up to help and aren't motivated by money.
CHRIS: That's very cool. When it comes to prototyping things with Integer 32, what kind of projects are people coming to you and asking you to prototype in Rust?
CAROL: A lot of them are existing projects that they have and written in something else that they want to either perform better and be safer as opposed to rewriting it in C or C++ to get performance out of it. Sometimes, they want something to interface with other Rust things. We’re starting to see projects like that but mostly, they have a hunch that Rust will be good for their projects and solve some problem they're having with their current implementation. We scope their projects way down to whatever will let them evaluate, whether Rust is a good fit or not and we go with that.
STEPHANIE: Going from there, the question that I have is why Rust? I don't know a lot about Rust so I'd like to know what would be some of the benefits of using Rust, if you're familiar with programming. If you are in web development like I'm familiar with Ember, why would I like to use Rust or learn Rust?
CAROL: I could talk all day about this. I really love working with Rust. I feel like it is adding more tools to help me to write better code and taking care of little details that usually I would have to spend a lot of brainpower thinking about to get right all the time. But now I can actually concentrate on whatever it is I'm trying to get done and let the compiler take care of those details for me. The way it's implemented, it happens really fast.
The way I got into Rust was I'm a Rails developer previously to this job and I spend a lot of time optimizing Rails, looking for places where essentially too many Ruby objects and memory leaks and [inaudible] a lot of trying to make Rails go faster. At some point, you can't. There’s only so far you can take Ruby and Rails so I look at where I want my career to go next and I love making things go faster but I'm terrified of C. I should be nowhere near production C.
You have to spend years learning all the quirks and all the ways that C can go wrong and crash and be insecure. Around this time, I know you had Steve Klabnik on the show, in the previous episode. Steve is from Pittsburgh, where I am and he came home for Christmas one year and came to a Ruby meet up and was talking about this new language called Rust and how awesome it was. Steve gets distracted by new awesome things all the time so I was like, "Yeah, yeah, okay, whatever."
The next year, he came home for Christmas again and was still talking about how awesome Rust was. At that point, I was like, "There's got to be something to this." At that point, he was writing his book, 'Rust for Rubyist' which has lead into his work on the Rust programming language book. I was like, "Rust for Rubyist? I can handle this. This is something I can do and capable of," so I started reading his book and submitting corrections and things which is again, how I got involved with the Rust programming language book.
But once you get that that's the part that enables your code to go faster without needing the garbage collector. You actually don't have to think about your memory management as much as you would in C, where you have to say, "Please give me some memory." Okay, I'm done with it now. You are manually managing your memory but you don't have to think about it as much because the compiler is thinking about it for you, if that makes sense.
In a relatively short time, it has developed a level of ergonomics that I'm envious of, even in the 'more comfortable' languages, things like Cargo, things like the compiler is really great but now the compiler has really friendly and informative error messages so that 'undefined is not a function' never happens but when you try to make it happen, it now shows you like, "No, no, no. On this exact line, in this place, this is where you're doing it wrong."
But I recently heard it and I'm curious if you know anything about it that there's also development on a Rust language service, kind of like I guess TypeScript test, where it's a whole set of tools that you can run under the hood that any editor can plug into so that you just get this tool box of things to help you develop in Rust that are all packaged up and handed over and all you have to do is hook into it. Have you try that at all? Are you familiar with that?
CAROL: I am not. I've been watching but I haven't worked on it and I haven't tried it out yet. I am excited for the language server because it's going to enable IDEs to do more interesting things. Coming from Ruby where it's so dynamic that you can't do things like ensure that you renamed all of the places and method it's called because you can't know that. I've read books like Michel Feathers' Working Effectively with Legacy Code and a lot of the chapters in that talked about leaning on your IDE, on your refactoring tools to do automated refactoring.
RubyMine has a few of those things but not all of them because it's just impossible so I'm really looking forward to having real refactoring tools that can let you do automated refactorings and things like that that are possible in other statically-typed languages but with Rust in an IDE. I haven't used an IDE in years because I haven't found them to be useful but once the language server is up and running, I'm thinking about going back to an IDE so it's definitely exciting.
CHRIS: That's some pretty cool right now. There's one called Clippy which I love because of the name like it takes you back to my Microsoft Word days. There’s a lot of very good stuff that they have added that I didn't expect from a 'systems language' but it has definitely benefited from a lot of things that people in the scripting world have learned.
I want anyone listening to this who thinks, "I can't do system programming. It was too hard." No, you totally can. You can do Rust. Rust is going to let you do this and that's why Rust is really exciting because it's enabling a whole new group of people to get into the systems programming space where things need to be optimized and faster and letting people build these sorts of things without having the programs be vulnerable to crashes and security bugs and things like that. It’s really, really exciting.
CHRIS: Very cool.
STEPHANIE: I'm curious in Rust, if there's an error, how would you know that there's an error? Is the whole thing going to stop? Is it going to break? Do you get a useful stack trace? What would I expect to see?
CAROL: A lot of the errors in Rust are at compile time. It won't even let you try to run your code if you have certain kinds of problems and they tried to move as much as possible into that compile time space. There are always going to be things that you can't catch a compile time like the user enters a number that's too big for whatever you're trying to do. That’s still going to be a runtime error because we can't possibly know what a user is going to put in when you're compiling. They've done a lot of work on the compiler errors as Chris was talking about, to make them friendlier and point here's where your error is, here's why it's happening, here's a hint as to how you might want to fix it. This has been really great.
I was volunteering in a local code school with students just starting Ruby and I'm used to Ruby's error messages by now but they were just getting started and getting all sorts of errors and I was like, "Wow, these error messages are not helpful at all," and I forgot how bad that is and how confusing it can be for a beginner to just think you understand, think you have got it working and then you go and run the code and it's just like 'string is not a symbol' or whatever. The worst is when you forget to close the block and just expected to see [inaudible] end of file instead and it's not helpful at all. I was just like apologizing the whole time like, "I'm sorry. This is telling you that you need to write 'end' at the end of the file," but it's not telling you that in any way you could possibly know that. That made me appreciate much more all of the work that's going into Rust error messages that are really trying to help.
Some people talk about, especially the borrow checker, fighting the borrow checker like they're not used to having a compiler tell them their code is wrong so often so people talk about fighting the borrow checker a lot but it's not trying to fight with you. It's not trying to make you feel bad about your code. It’s trying to help you make your code better and prevent errors that might happen at runtime by catching them earlier. I actually have a little project called Rustlings that is full of little snippets of Rust that intentionally don't compile. You run them and you get an error message right away and your job is to read the error message and learn how to fix it.
When I was starting out, I was really frustrated because I was trying to do something and I would get an error message and I would have to stop whatever is doing to deal with the error message. I was like, "If I could just get some practice just dealing with the error messages and learning how to fix them so that when I'm trying to do something else, I already have experience fixing that kind of error," so that's how that project came to be and people found that useful. I haven't had much time to work on it lately but it could definitely use more examples because I think people are used to error messages that are not helping. People used to back traces that are really long and don't say anything useful.
Sometimes, you don't stop and read and think but the Rust error messages are really trying to help you and often times, they are telling you exactly what you need to do to change the code to work. I think getting practice seeing the compiler as more of a pair who is trying to help you and not someone who's trying to reject all your code is a different mindset that I don't think people are used to but I think it's really useful.
STEPHANIE: That's excellent. I was going to ask you if there are any resources or any repos to check out for someone who is interested in getting into Rust. It's funny, last night I was poking around with Python and there's something similar to Rustlings. It's called Python Koans and it's basically like what you're already familiar if you do web development. You want to get your test to pass so it'll tell you, you need to think about this one or you need to meditate on this and then you try to get it to pass and then it says you have reached enlightenment or you have not yet reached enlightenment and you have to sit there and think about it and then run it again. It’s very useful in trying to get started with language in a way that you are already sort of familiar with.
CAROL: Yeah, I've definitely gotten inspiration from the Koans project that have existed in other languages. There's also an exercism track for Rust that people found really useful and of course, I'm working with Steve Klabnik on the Rust programming language book. We're rewriting the whole thing so there's an existing version that if you go to the Rust documentation and click on book, you'll get the existing version which is complete but the new version is going to be way better. Especially with the explanations of ownership and borrowing, people have said that the new version is way, way better than the old version.
Someone even made the analogy of doing medical research and you see that trial case is doing so much better than the placebo case that is not ethical to continue the trial. It's more ethical to stop the trial and use the new thing because it's helping so many people. Someone was like, "You need to replace the old book with the new book right now because it's so much better," but the new book isn't complete yet.
The new book is in a different repo which we can put in the show notes so I'd recommend starting with the new book and then working back and forth with the old book once you run out of content. But we're getting closer all the time so hopefully, that should be done and printed by No Starch sometime in 2017 --
CHRIS: Woah! It's being printed by No Starch?
CHRIS: That's cool. I didn't know that. Congratulations.
CAROL: I thought Steve mentioned that in the last one.
CHRIS: He may have but he talked about it a long time ago and I thought he always meant the old one. How long have you been rewriting the Rust book for?
CAROL: It's been a while.
CHRIS: Longer than I knew about then, probably.
CAROL: It's kind of like software. It’s more work than you think it's going to be and estimating, it's going to be done when it's done. If you kept telling people, "It's going to be out on this time," and like Steve, "There's no way it's not getting done by then," so now he's not allowed to say it when it's going to be out.
CAROL: I'm really grateful to see this opportunity because I don't think I would have written a book on my own and I'm learning a whole lot about the process of writing a book. It turns out there's a lot of editing, a lot of back and forth, a lot of trying to build a narrative through this long stretch of text so that you're building on top of what you've already covered and not introducing things that aren't mentioned. It's a lot of work and I'm learning a lot and I have no idea when it's going to be [inaudible] because I think there's more work that I still don't know about coming, as we get closer to going to print. It’s definitely one of those things that you can't make agile because you've got to put it on paper that costs money and it's going to be around a long time at some point. It’s definitely a different kind of working that I'm used to with software.
CHRIS: Although, I have to say, I clicked around and I think this is the new version: Rustlings.Github.io/Book. Is that the new one?
CAROL: Yes, that's the new version.
CHRIS: There is a lot here and it's not quite what I would have expected to see here like it's not done yet. I've been clicking links and I have yet to find one that says 'to-do'.
CAROL: I think 15 through 20 are like outlines right now. We’re maybe three-quarters through with the content but then we have to go through revisions and editing and copyediting.
CHRIS: I'm looking at the headings and I was a big fan of the first Rust book but I can already see it calling out things I wish had been hit on more specifically in the original book. There’s a lot of good looking stuff here so I'm excited about this. I'm going to go and read this thing.
CAROL: I'm excited for people to read it.
CHRIS: Earlier, you were talking about one of the things that is really nice about the Rust tooling is that Cargo makes it really easy to bring in dependencies. I happen to know that you are recently, I believe the maintainer of Crates.io which is where all of Cargoes crates, which are the libraries are hosted. Is that correct?
CAROL: That is correct. I have commitment Crates.io now which is very exciting. Crates.io is like Ruby Gems or npm. It’s the site where people publish their libraries and you can go and search for a library for what you need. As part of the Rust 2017 goals, we want to make it easier for people to find high-quality libraries that do the things they needed to do. I've been doing some work on adding badges and categories. Rust makes major decisions on the language and on things through an RFC process, which I think Ember is doing now too. I forget which way we stole that. Do we steal it from Ember or did you steal from us? I can't remember.
CHRIS: If I remember right, I think -- I could be wrong, Twitter -- Ember did it first. Rust borrowed it and then added the 'how do we teach this?' section. I think Ember took that back and added it to their RFCs.
CAROL: Okay, I'm super excited about that section. Now, when you propose a change of language, you have to go through this RFC process where you write up what you want to change, why you want to change it, any downsides, any alternative designs. Then the community talks about it and makes comments when you revise it and things like that. Now, there's a new section that just got added. That's 'how do we teach this?' Before something can be stabilized in the language, you have to document it. This is still kind of starting to take effect but I'm super excited about it because people can't use something unless they know how to use it.
Right now, Steve's the only person getting paid full time to work on the documentation and I need him to write the book so I'm excited that more people will be thinking about documentation and thinking about how to help people use their new features. Anyway, I have an RFC about how to rank Crates within a category that we're trying to work through. In some automated ways, we can recommend different Crates for different purposes. I'm working through that with the community to try and figure out how to best recommend Crates in different circumstances.
Crates.io is written in Rust and it performs really well. It just got added to the Heroku things so you can deploy it too. Looking at the analytics and their response times for is just like the Ruby apps I work on would be thrilled to have these stats. The backend of it is Rust, the frontend is Ember and [inaudible] who was an Ember person is also interested in Rust and he thinks Rust on the backend and Ember on the frontend worked really well together. He’s always trying to figure out ways that we can work together.
CHRIS: Very cool. I'm personally very interested in that and will likely hit you up. But I'm sure many of our listeners will as well because I think we have a lot of Ember-friendly listeners so look Carol up because it sounds like she could use some help. Actually, I'm curious, the backend, I know that pretty recently, Rust has kind of gone through this period of kind of explosion in terms of Rust as a web language. There have been a number of different things that have come out pretty recently for a web framework in Rust or there's that Tokio thing. I know Diesel is like the ORM for Rust in talking to databases. It looks like it's about to hit 1.0. There's a lot of stuff happening so I'm curious, what are you using to write the backend. I know you're using Rust but are you using one of these frameworks or have you rolled your own? How's that work right now?
CAROL: Crates.io is one of the first web apps that has been written in Rust. Actually, if you look at the backend code, you'll see SQL being built by hand. It’s going to the Rust postgres library so it has SQL injection protection. All the things are [inaudible] so don't worry about that but they're still like SQL with the Rust code so it's not using an ORM yet. I'm going to have to look up. There is a library that is using that I'm blanking on the name of it for but it's very low level. It just let you send HTTP requests and let you respond to HTTP.
We’re in kind of a Cambrian explosion period with Rust web frameworks. There’s a lot of different ones. One that I'm excited about that I haven't gotten to tryout yet is Rocket. That was just released. The thing I love about Rocket is that everyone's really excited about it because it was announced and they have an awesome website with lots of awesome docs so that should be a lesson to any open source project that's launching is if you want to get excited about it, you've got to launch some docs. That will help so much.
There's a lot of different frameworks happening. They’re still little trilobites and little animals that can't walk on land on their own quite yet so there's still no Rails. There's the pieces of Rails. There’s Diesel which would act like a record. There's Nickel and Pencil and Iron and Rocket. Tokio is the async framework that is getting more and more stable by the day. We got to try it out on a project recently and it's pretty fun. I still am working on wrapping my head around promises and futures and working in that way but I think as that stabilizes and people use that, that is going to cause like another explosion of libraries that enable really fast but safe web backend stuff, which I think is really exciting.
If you're looking for the Rails experience of being able to plug things together and nicely, just declare a few things, it works but not quite there yet. But if it excites you to try out new things and figure out the best ways to do the things you want to do in Rust, this is a great time to jump in and help.
CHRIS: I will say the Rocket website is beautiful and it even has this templating section, a testing library section. This is very exciting. It really looks like as the closest thing to a Rail-style web framework that I've seen in Rust so far. People should definitely check this stuff out. I'm curious, I know a lot is really interested in Rust and Ember, which doesn't surprise me because lots is really interested in Ember in general, which I think is awesome. But is there anything specific about working with the Rust and Ember together that seems, especially well suited or even like some gotchas that you guys have run into?
One of the things I'm thinking of is like Ember is really big into JSON API spec and I don't know if Rust has a JSON API library for serializing things in that format. Is that something you guys have to tackle at all?
CAROL: There might be. I'm not sure. Crates.io is using the Rust API adapter for Ember so we might not be keeping up with the latest of Ember. But I know there are people who want them to interface them better with each other. Actually, that's an interesting thing. Both Ember and Rust are on a six-week release trains sort of things so the way Rust people will say -- I don't know if Ember people do -- is stability without stagnation so they're both changing.
Rust has backwards-compatibility guarantees so the code you wrote with Rust 1.0 is still going to compile today. You might have some warnings and there's probably new cooler stuff that you could switch to but it's still going to compile. I'm not sure about Ember's upgrade path things. Someone just sent in a pull request that we merged like three days ago to upgrade us from two Ember point versions. There were a couple of things that like [inaudible] and we weren't doing quite right and we had to fix. It’s been interesting to kind of fit together, keep both of the sides, update it and upgrade it and continuing on using the best things. But I think they have similar philosophies around making things better all the time.
CHRIS: Yeah, the whole stable upgrade path and backwards-compatibility guarantees is definitely mirrored on the Ember side of things. I can see that being a little kind of comforting place to be knowing that both your frontend and your backend are not going to suddenly just break on you one day because some new feature came out that breaks your router or something. That’s very cool.
One of the thing that I know you're involved in -- you're involved in a lot of things -- when it comes to Rust, it's very cool. But you also run or a co-run a conference, right? Rust Belt Rust?
CAROL: Yeah, we had our first year in 2016 in Pittsburgh. I ran Steel City Ruby before then so I love running conferences and I love having them near me one, because it's convenient and I get to trick all of my friends into coming to visit me. But two, because there's a lot of tech stuff happening in the Rust Belt and places that aren't San Francisco or New York. People don't necessarily know about that and people who live here don't necessarily have the opportunities to travel as easy to conferences. I sort of start Rust Belt Rust, one because of the pun opportunity and one of our speakers drew a little bar graph. There were three conferences last year. There was Rust Fest in Europe which has [inaudible] amount of Rust. There’s RustConf, the official Rust Conference in Portland that has a lot amount of Rust and then Rust Belt Rust has double the amount of Rust in its name so we're the Rust-serious Rust conference.
We're going to do it again, in 2017 we're going to move it to a different Rust Belt city. I'm not going to say which one yet but we're closing in on a date and a venue in the Rust Belt city so watch out for an announcement on that. It was a lot of fun. We had a day of workshops and then a day of single-track talks and a lot of time for conversation. A bunch of the core team members came out and it was fun talking with a friend of mine who was trying out something with Tokio. This was in October so Tokio was still working towards their first big release and he was trying to do something with Tokio. I looked over and I saw Carl Lerche, Alex Crichton and Aaron Turon standing together and talking like 30 feet from us and I was like, "If only the three people working on Tokio were nearby to answer your question --" so he just walked over and talked about Tokio with them.
I love getting people together to talk to other people working with things, talk to the people who are working on the things they're using and meeting the people behind the names on the internet so I love running conferences and having events like that.
STEPHANIE: Carol, you have a Rust consultancy called Integer 32. How is that going?
CAROL: It's going pretty well. We’re learning a lot. One of the reasons I wanted to start it is because I felt like I wasn't learning more in my job. In my Rails job, I felt like I had kind of tapped out with that knowledge. In starting a business, I get to learn a lot of stuff like sales and marketing and taxes and invoices. Sometimes, I even get to program a little. We're still learning how to effectively find our target customers. We do have availability, if anyone listening is interested in hiring some Rust experts.
Right now, I'm trying to figure out when can we bring more people on the team. I'm trying to decide if we can have an intern for the summer. It should be fun so yeah, it's going pretty well. It's been a slow build but we're lucky enough to have savings and be able to spend some time building our business but it's been really gratifying to feel like I'm in charge of my destiny somewhat, as opposed to the whims of a company.
STEPHANIE: And if I were interested in some Rust consulting, what would be the best way to reach you?
CAROL: We have a website at Integer32.com and a contact form on there.
STEPHANIE: Thank you so much for speaking with us, Carol. It was a pleasure. I feel like I learned a lot about Rust.
CAROL: Thank you for having me.
STEPHANIE: All right, y'all. That’s it from us. Thank you so much for tuning in. Until next time. Bye-bye.