367: Value Objects

Joël's been traveling. Stephanie's working on professional development. She's also keeping up a little bit more with Ruby news and community news in general and saw that Ruby 3.2 introduced a new class called data to its core library for the use case of creating simple value objects. This episode is brought to you by Airbrake. Visit Frictionless error monitoring and performance insight for your app stack. Maggie Appleton's Tools for Thought Episode on note-taking with Amanda Beiner Obsidian Zettelkasten Evergreen notes New Data class Joël's article on value objects Episode on specialized vocabulary Primitive Obsession Transcript: AD: thoughtbot is thrilled to announce our own incubator launching this year. If you are a non-technical founding team with a business idea that involves a web or mobile app, we encourage you to apply for our eight-week program. We'll help you move forward with confidence in your team, your product vision, and a roadmap for getting you there. Learn more and apply at tbot.io/incubator. STEPHANIE: Hello and welcome to another episode of The Bike Shed, a weekly podcast from your friends at thoughtbot about developing great software. I'm Stephanie Minn. JOËL: And I'm Joël Quenneville. And together, we're here to share a little bit of what we've learned along the way. STEPHANIE: So, Joël, what's new in your world? JOËL: I've been traveling for the past few weeks in Europe. I just recently got back to the U.S. and have just gotten used to drinking American-style drip coffee again after having espresso every day for a few weeks. And it's been an adjustment. STEPHANIE: I bet. I think that it's such a downgrade compared to European espresso. I remember when I was in Italy, I also would really enjoy espresso every day at a local cafe and just be like sitting outside drinking it. And it was very delightful. JOËL: They're very different experiences. I have to say I do enjoy just holding a hot mug and sort of sipping on it for a long time. It's also a lot weaker. You wouldn't want to do a full hot mug of espresso. That would just be way too intense. But yeah, I think both experiences are enjoyable. They're just different. STEPHANIE: Yeah. So, that first day with your measly drip coffee and your jet lag, how are you doing on your first day back at work? JOËL: I did pretty good. I think part of the fun of coming back to the U.S. from Europe is that the jet lag makes me a very productive morning person for a week. Normally, I'm a little bit more of an evening person. So I get to get a bit of an alter ego for a week, and that helps me to transition back into work. STEPHANIE: Nice. JOËL: So you've also been on break and have started work again. How are you feeling productivity-wise, kicking off the New Year? STEPHANIE: I'm actually unbooked this week and the last week too. So I'm not working on client projects, but I am having a lot of time to work on just professional development. And usually, during this downtime, I also like to reassess just how I'm working, and lately, what that has meant for me is changing my note-taking process. And I'm really excited to share this with you because I know that you have talked about this on the show before, I think in a previous episode with a guest, Amanda Beiner. And I listened to that episode, and I was really inspired because I was feeling like I didn't have a note-taking system that worked super well for me. But you all talked about some tools you used and some, I guess, philosophies around note-taking that like I said, I was really inspired by. And so I hopped on board the Obsidian train. And I'm really excited to share with you my experience with it. So I really like it because I previously was taking notes in my editor under the impression that, oh, like, everything is in one place. It'll be like a seamless transition from code to note-taking. And I was already writing in Markdown. But I actually didn't like it that much because I found it kind of distracting to have code things kind of around. And if I was navigating files or something, something work or code-related might come up, and that ended up being a bit distracting for me. But I know that that works really well for some people; a coworker of ours, Aji, I know that he takes his notes in Vim and has a really fancy setup for that. And so I thought maybe that's what I wanted, but it turns out that what I wanted was actually more of a boundary between code and notes. And so, I was assessing different note-taking and knowledge management software. And I have been really enjoying Obsidian because it also has quite a bit of community support. So I've installed a few plugins for just quality-of-life features like snippets which I had in my editor, and now I get to have in Obsidian. I also installed things like Natural Language Dates. So for my running to-do list, I can just do a shortcut for today, and it'll autofill today's date, which, I don't know, because for me, [laughs] that is just a little bit less mental work that I have to do to remember the date. And yeah, I've been really liking it. I haven't even fully explored backlinking, and that connectivity aspect, which I know is a core feature, but it's been working well for me so far. JOËL: That's really exciting. I love notes and note-taking and the ways that we can use those to make our lives better as developers and as human beings. Do you have a particular system or way you've approached that? Because I know for me, I probably looked at Obsidian for six months before I kind of had the courage to download it because I didn't want to go into it and not have a way to organize things. I was like; I don't want to just throw random notes in here. I want to have a system. That might just be me. But did you just kind of jump into it and see, like, oh, a system will emerge? Did you have a particular philosophy going in? How are you approaching taking notes there? STEPHANIE: That's definitely a you thing because I've definitely had the opposite experience [laughs] where I'm just like, oh, I've downloaded this thing. I'm going to start typing notes and see what happens. I have never really had a good organizational system, which I think is fine for me. I was really leaning on pen and paper notes for a while, and I still have a certain use case for them. Because I find that when I'm in meetings or one-on-ones and taking notes, I don't actually like to have my hands on the keyboard because of distractions. Like I mentioned earlier, it's really easy for me to, like, oh, accidentally Command-Tab and open Slack and be like, oh, someone posted something new in Slack; let me go read this. And I'm not giving the meeting or the person I'm talking to my full attention, and I really didn't like that. So I still do pen and paper for things where I want to make sure that I'm not getting distracted. And then, I will transfer any gems from those notes to Obsidian if I find that they are worth putting in a place where I do have a little bit more discoverability and eventually maybe kind of adding on to my process of using those backlinks and connecting thoughts like that. So, so far, it's truly just a list of separate little pages of notes, and yeah, we'll see how it goes. I'm curious what your system for organizing is or if you have kind of figured out something that works well for you. JOËL: So my approach focuses very heavily on the backlinks. It's loosely inspired by two similar systems of organization called Zettelkasten and evergreen notes. The idea is that you create notes that are ideas. Typically, the title is like a thesis statement, and you keep them very short, focused on a single thing. And if you have a more complex idea, it probably breaks down into two or three, and then you link them to each other as makes sense. So you create a web of these atomic ideas that are highly interconnected with each other. And then later on, because I use this a lot for either creating content in the future or to help refine my thinking on various software topics, so later on, I can go through and maybe connect three or four things I didn't realize connected together. Or if I'm writing an article or a talk, maybe find three or four of these ideas that I generated at very different moments, but now they're connected. And I can make an article or a talk out of them. So that's sort of the purpose that I use them for and how I've organized things for myself. STEPHANIE: I think that's a really interesting topic because while I was assessing different software for note-taking and, like I said, knowledge management, I discovered this blog post by Maggie Appleton that was super interesting because she is talking about the term tools of thought which a lot of these different software kind of leveraged in their marketing copy as like, oh, this software will be like the key to evolving your thinking and help you expand making connections, like you mentioned, in ways that you weren't able to before. And was very obviously trying to upsell you on this product, and she -- JOËL: It's over the top. STEPHANIE: A little bit, a little bit. So in this article, I liked that she took a critical lens to that idea and rooted her article in history and gave examples of a bunch of different things in human history that also evolved the ways humans were able to express their thoughts and solve problems. And so some of the ones that she listed were like storytelling and oral tradition. Literally, the written language obviously [laughs] empowered humans to be able to communicate and think in ways that we never were before but also drawings, and maps, and spreadsheets. So I thought that was really cool because she was basically saying that tools of thought don't need to be digital, and people claiming that these software, you know, are the new way to think or whatever, it's like, the way we're thinking now, but we also have this long history of using and developing different things that helped us communicate with each other and think about stuff. JOËL: I think that's something that appealed to me when I was looking at some of these note-taking systems. Zettelkasten, in particular, predates digital technology. The original system was built on note cards, and the digital stuff just made it a little bit easier. But I think also when I was reading about these ideas of keeping ideas small and linking them together, I realized that's already kind of how I tend to organize information when I just hold it in my brain or even when I try to do something like a tweet thread on Twitter where I'll try to break it up. It might be a larger, more complex idea, but each tweet, I try to get it to kind of stand on its own to make it easier to retweet and all that. And so it becomes a chain of related ideas that maybe build up to something, but each idea stands on its own. And that's kind of how in these systems notes end up working. And they're in a way that you can kind of remix them with each other. So it's not just a linear chain like you would have on Twitter. STEPHANIE: Yeah, I remember you all in that episode about note-taking with Amanda talked about the value of having an atomic piece of information in every note that you write. And since then, I've been trying to do that more because, especially when I was doing pen and paper, I would just write very loose, messy thoughts down. And I would just think that maybe I would come back to them one day and try to figure out, like, oh, what did I say here, and can I apply it to something? But it's kind of like doing any kind of refactoring or whatever. It's like, in that moment, you have the most context about what you just wrote down or created. And so I've been a little more intentional about trying to take that thought to its logical end, and then hopefully, it will provide value later. What you were saying about the connectivity I also wanted to kind of touch on a little bit further because I've realized that for me, a lot of the connection-making happens during times where I'm not very actively trying to think, or reflect, or do a lot of deep work, if you will. Because lately, I've been having a lot of revelations in the shower, or while I'm trying to fall asleep, or just other kinds of meditative activity. And I'm just coming to terms with that's just how my brain works. And doing those kinds of activities has value for me because it's like something is clearly going on in my brain. And I definitely want to just honor that's how it works for me. JOËL: I had a great conversation recently with another colleague about the gift of boredom and how that can impact our work and what we think about, and our creativity. That was really great. Sometimes it's important to give ourselves a little bit more blank space in our lives. And counter-intuitively, it can make us more productive, even though we're not scheduling ourselves to be productive. STEPHANIE: Yes, I wholeheartedly agree with that. I think a lot about the feeling of boredom, and for me, that is like the middle of summer break when you're still in school and you just had no obligations whatsoever. And you could just do whatever you wanted and could just laze around and be bored. But letting your mind wander during those times is something I really miss. And sometimes, when I do experience that feeling, I get a little bit anxious. I'm like, oh, I could be doing something else. There's whatever endless list of chores or things that are, quote, unquote, "productive." But yeah, I really like how you mentioned that there is value in that experience, and it can feel really indulgent, but that can be good too. MID-ROLL AD: Debugging errors can be a developer’s worst nightmare...but it doesn’t have to be. Airbrake is an award-winning error monitoring, performance, and deployment tracking tool created by developers for developers that can actually help cut your debugging time in half. So why do developers love Airbrake? It has all of the information that web developers need to monitor their application - including error management, performance insights, and deploy tracking! Airbrake’s debugging tool catches all of your project errors, intelligently groups them, and points you to the issue in the code so you can quickly fix the bug before customers are impacted. In addition to stellar error monitoring, Airbrake’s lightweight APM helps developers to track the performance and availability of their application through metrics like HTTP requests, response times, error occurrences, and user satisfaction. Finally, Airbrake Deploy Tracking helps developers track trends, fix bad deploys, and improve code quality. Since 2008, Airbrake has been a staple in the Ruby community and has grown to cover all major programming languages. Airbrake seamlessly integrates with your favorite apps to include modern features like single sign-on and SDK-based installation. From testing to production, Airbrake notifiers have your back. Your time is valuable, so why waste it combing through logs, waiting for user reports, or retrofitting other tools to monitor your application? You literally have nothing to lose. Head on over to airbrake.io/try/bikeshed to create your FREE developer account today! JOËL: So you mentioned recently that you've had a lot of revelations or new ideas that have come upon you or that you've been able to dig into a little bit more. Is there one you'd like to share with the audience? STEPHANIE: Yeah. So during this downtime that I've had not working on client work, I have been able to keep up a little bit more with Ruby news or just community news in general. And in, I think, an edition of Ruby Weekly, I saw that Ruby 3.2 introduced this new class called data to its core library for the use case of creating simple value objects. And I was really excited about this new feature because I remembered that you had written a thoughtbot blog post about value objects back in the summer that I had reviewed. That was an opportunity that I could make a connection between something happening in recent news with some thoughts that I had about this topic a few months ago. But basically, this new class can be used over something like a struct to create objects that are immutable in their values, which is a big improvement if you are trying to follow value objects semantics. JOËL: So, I have not played around with the new data class. How is it different from the existing struct that we have in Ruby? STEPHANIE: So I think I might actually answer that first by saying how they're similar, which is that they are both vehicles for holding pieces of data. So we've, in the past, been able to use a struct to very cheaply and easily create a new class that has attributes. But one pitfall of using a struct when you're trying to implement something like a value object is that structs also came with writer methods for all of its members. And so you could change the value of a member, and that it kind of inherently goes against the semantics of a value object because, ideally, they're immutable. And so, with the data class, it doesn't offer writer methods essentially. And I think that it freezes the instance as well in the constructor. And so even if you tried to add writer methods, you would eventually get an error. JOËL: That's really convenient. I think that may be an area where I've been a little bit frustrated with structs in the past, which is that they can be modified. They basically get treated as if they're hashes with a slightly nicer syntax to interact with them. And I want slightly harder boundaries around the data. Particularly when I'm using them as value objects, I generally don't want people to modify them because that might lead to some weird bugs in the code where you've got a, I don't know, something represents a time value or a date value or something, and you're trying to do math on it. And instead of giving you a new time or date, value just modifies the first one. And so now your start date is in the past or something because you happen to subtract a time from it to do a calculation. And you can't assign it to a variable anywhere. STEPHANIE: Yeah, for sure. Another kind of pitfall I remember noticing about structs were that the struct class includes the enumerable module, which makes a struct kind of like a collection. Whereas if you are using it for a value object, that's maybe not what you want. So there was a bit of discourse about whether or not the data class should inherit from struct. And I think they landed on it not inheriting because then you can draw a line in the sand and have that stricter enforcement of saying like, this is what a data as value object should be, and this is what it should not be. So I found that pretty valuable too. JOËL: I think I've heard people talk about sort of two classes of problems that are typically solved with a struct; one is something like a value object where you probably don't want it to be writable. You probably don't want it to be enumerable. And it sounds like data now takes on that role very nicely. The other category of problem is that you have just a hash, and you're trying to incrementally migrate it over to some nicer objects in some kind of domain. And struct actually gives you this really nice intermediate phase where it still mostly behaves like a hash if you needed to, but it also behaves like an object. And it can help you incrementally transition away from just a giant hash into something that's a little bit more programmatic. STEPHANIE: Yeah, that's a really good point. I think struct will still be a very viable option for that second category that you described. But having this new data class could be a good middle ground before you extract something into its own class because it better encapsulates the idea of a value object. And one thing that I remember was really interesting about the article that you wrote was that sometimes people forget to implement certain methods when they're writing their own custom value objects. And these come a bit more out of the box with data and just provide a bit more like...what's the word I'm looking for? I'm looking for...you know when you're bowling, and you have those bumpers, I guess? [laughs] JOËL: Uh-huh. STEPHANIE: They provide just like safeguards, I guess, for following semantics around value objects that I thought was really important because it's creating an artifact for this concept that didn't exist. JOËL: And to recap for the audience here, the difference is in how objects are compared for equality. So value objects, if they have the same internal value, even if they're separate objects in memory, should be considered equal. That's how numbers work. That's how hashes work. Generally, primitives in Ruby behave this way. And structs behave that way, and the new data class, it sounds, also behaves that way. Whereas regular objects that you would make they compare based off of the identity of the object, not its value. So if you create two user instances, not ActiveRecord, but you could create a user class, you create two instances in memory. They both have the same attributes. They will be considered not equal to each other because they're not the same instance in memory, and that's fine for something more complex. But when you're dealing with value objects, it's important that two objects that represent the same thing, like a particular time for a unit of measure or something like that, if they have the same internal value, they must be the same. STEPHANIE: Right. So prior to the introduction of this class, that wasn't really enforced or codified anywhere. It was something that if you knew what a value object was, you could apply that concept to your code and make sure that the code you wrote was semantically aligned with this concept. And what was kind of exciting to me about the addition of this to the core class library in Ruby is that someone could discover this without having to know what a value object is like more formally. They might be able to see the use of a data class and be like, oh, let me look this up in the official Ruby docs. And then they could learn like, okay, here's what that means, and here's some rules for this concept in a way that, like I mentioned earlier, felt very implicit to me prior. So that, I don't know, was a really exciting new development in my eyes. JOËL: One of the first episodes that you and I recorded together was about the value of specific vocabulary. And I think part of what the Ruby team has done here is they've taken an implicit concept and given it a name. It's extracted, and it has a name now. And if you use it now, it's because you're doing this data thing, this value object thing. And now there's a documentation page. You can Google it. You can find it rather than just be wondering like, oh, why did someone use a struct in this way and not realize there are some implicit semantics that are different? Or wondering why did the override double equals on this custom class? STEPHANIE: Yeah, exactly. I think that the introduction of this class also provides a solution for something that you mentioned in that blog post, which was the idea of testing value objects. Because previously, when you did have to make sure that you implemented methods, those comparison methods to align with the concept of a value object, it was very easy to forget or just not know. And so you provided a potential solution of testing value objects via an RSpec shared example. And I remember thinking like, ooh, that was a really hot topic because we had also been debating about shared examples in general. But yeah, I was just thinking that now that it's part of the core library, I think, in some ways, that eliminates the need to test something that is using a data class anyway because we can rely a little bit more on that dependency. JOËL: Right? It's the built-in behavior now. Do you have any fun uses for value objects recently? STEPHANIE: I have not necessarily had to implement my own recently. But I do think that the next time I work with one or the next time I think that I might want to have something like a value object it will be a lot easier. And I'm just excited to play around with this and see how it will help solve any problem that might come up. So, Joël, do you have any ideas about when you might reach for a data object? JOËL: A lot of situations, I think, when you see the primitive obsession smell are a great use case for value objects, or maybe we should call them data objects now, now that this is part of Ruby's vocabulary. I think I often tend to; preemptively sounds bad, but a lot of times, I will try to be careful. Anytime I'm doing anything with raw numbers, magic strings, things like that, I'll try to encapsulate them into some sort of struct. Or even if it's like a pair of numbers, it always goes together, maybe a latitude and longitude. Now, those are a pair. Do I want to just be passing around a two-element array all the time or a hash that would probably make a very nice data object? If I have a unit of measure, some number that represents not just the abstract concept of three but specifically three miles or three minutes, then I might reach for something like a data class. STEPHANIE: Yeah, I think that's also true if you're doing any kind of arithmetic or, in general, trying to compare anything about two of the same things. That might be a good indicator as well that you could use something richer, like a value object, to make some of that code more readable, and you get some of those convenient methods for doing those comparisons. JOËL: Have you ever written code where you just have like some number in the code, and there's a comment afterwards that's like minutes or miles or something like that, just giving you the unit as a comment afterwards? STEPHANIE: Oh yeah. I've definitely seen some of that code. And yeah, I mean, now that you mentioned it, that's a great use case for what we're talking about, and it's definitely a code smell. JOËL: It can often be nice as you make these more domain concepts; maybe they start as a data object, but then they might grow with their own custom methods. And maybe you extend data the same way you could extend a struct, or maybe you create a custom class to the point where the user...whoever calls that object, doesn't really need to know or care about the particular unit, just like when you have duration value. If you have a duration object, you can do the math you want. You can do all the operations and don't have to know whether it is in milliseconds, or seconds, or minutes because it knows that internally and keeps all of the math straight as opposed to just holding on to what I've done before, which is you have some really big number somewhere. You have start is, or length is equal to some big number and then comment milliseconds. And then, hopefully, whoever does math on that number later remembers to do the division by 1,000 or whatever they need. STEPHANIE: I've certainly worked on code where we've tolerated those magic numbers for probably longer than we should have because maybe we did have the shared understanding that that value represents minutes or milliseconds or whatever, and that was just part of the domain knowledge. But you're right, like when you see them, and without a very clear label, all of that stuff is implied and is really not very friendly for someone coming along in the future. As well as, like you mentioned earlier, if you have to do math on it later to convert it to something else, that is also a red flag that you could use some kind of abstraction or something to represent this concept at a higher level but also be extensible to different forms, so a duration to represent different amounts of time or money to represent different values and different currencies, stuff like that. JOËL: Do you have a guideline that you follow as to when something starts being worth extracting into some kind of data object? STEPHANIE: I don't know if I have particularly clear guidelines, but I do remember feeling frustrated when I've had to test really complicated hashes or just primitives that are holding a lot of different pieces of information in a way that just is very unwieldy when you do have to write a test for it. And if those things were encapsulated in methods, that would have been a lot easier. And so I think that is a bit of a signal for me. Do you have any other guidelines or gut instincts around that? JOËL: We mentioned the comment that is the unit. That's probably a...I wasn't sure if I would have to call it a code smell, but I'm going to call it a code smell that tells you maybe you should...that value wants to be something a little bit more than just a number. I've gotten suspicious of just raw integers in general, not enough to say that I'm going to make all integers data objects now, but enough to make me pause and think a lot of times. What does this number represent? Should it be a data object? I think I also tend to default to try to do something like a data object when I'm dealing with API responses. You were talking about hashes and how they can be annoying to test. But also, when you're dealing with data coming back from a third-party API, a giant nested hash is not the most convenient thing to work with, both for the implementation but then also just for the readability of your code. I often try to have almost like a translation layer where very quickly I take the payload from a third-party service and turn it into some kind of object. STEPHANIE: Yeah, I think the data class docs itself has an example of using it for HTTP responses because I think the particular implementation doesn't even require it to have attributes. And so you can use it to just label something rather than requiring a value for it. JOËL: And that is one thing that is nice about something like a data object versus a hash is that a hash could have literally anything in it. And to a certain extent, a data object is self-documenting. So if I want to know I've gotten to a shopping cart object from a third-party API, what can I get out of the shopping cart? I can look at the data object. I can open the class and see here are the methods I can call. If it's just a hash, well, I guess I can try to either find the documentation for the API or try to make a real request and then inspect the hash at runtime. But there's not really any way to find out without actually executing the code. STEPHANIE: Yeah, that's totally fair. And what you said about self-documenting makes a lot of sense. And it's always preferable than that stray comment in the code. [laughs] JOËL: I'm really excited to use the data class in future Ruby 3.2 projects. So I'm really glad that you brought it up. I've not tried it myself, but I'm excited to use it in future projects. STEPHANIE: On that note, shall we wrap up? JOËL: Let's wrap up. STEPHANIE: Show notes for this episode can be found at bikeshed.fm. JOËL: This show has been produced and edited by Mandy Moore. STEPHANIE: If you enjoyed listening, one really easy way to support the show is to leave us a quick rating or even a review in iTunes. It really helps other folks find the show. JOËL: If you have any feedback for this or any of our other episodes, you can reach us @_bikeshed, or you can reach me @joelquen on Twitter. STEPHANIE: Or reach both of us at [email protected] via email. JOËL: Thanks so much for listening to The Bike Shed, and we'll see you next week. ALL: Byeeeeeeeeeeeeee!!!!!!!!!!!! ANNOUNCER: This podcast is brought to you by thoughtbot, your expert strategy, design, development, and product management partner. We bring digital products from idea to success and teach you how because we care. Learn more at thoughtbot.com.Sponsored By:Airbrake: Deploy fearlessly and fix bugs faster with Airbrake Error & Performance Monitoring. Airbrake notifiers are available for all major programming languages and frameworks, and install in minutes, with an open-source SDK-based install and near-zero technical debt. Spend less time tracking down bugs and more time developing. Visit Frictionless error monitoring and performance insight for your app stack.Support The Bike Shed