Sveriges mest populära poddar

The Bike Shed

390: The Truth about Truthiness

40 min • 27 juni 2023

Joël's new work project involves tricky date formats. Stephanie has been working with former Bike Shed host Steph Viccari and loved her peer review feedback.

The concept of truthiness is tough to grasp sometimes, and JavaScript and Ruby differ in their implementation of truthiness.

  • Is this a problem?
  • Do you prefer one model over the other?
  • What can we learn about these design decisions?
  • How can we avoid common pitfalls?

Transcript:

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: So I'm on a new project at work. And I'm doing some really interesting work where I'm connecting to a remote database third-party system directly and pulling data from that database into our system, so not via some kind of API. And one thing that's been really kind of tricky to work with are the date formats on this third-party database.

STEPHANIE: Is the date being stored in an unexpected format or something like that?

JOËL: Yes. So there's a few things that are weird with it. So this is a value that represents a point in time, and it's not stored as a date-time value. Instead, it's stored separately as a date column and a time column. So a little bit of weirdness there. We can work with it, except that the time column isn't actually a time value. It is an integer.

STEPHANIE: Oh no.

JOËL: Yeah. And if you're thinking, oh, okay, an integer, it's going to be milliseconds since midnight or something like that, which is basically how Postgres' time of day works under the hood, nope, that's not how it works. It's a positional digit thing. So, if you've got the number, you know, 1040, that means 10:40 a.m.

STEPHANIE: Oh my gosh. Is this in military time or something like that, at least?

JOËL: Yes, it is military time. But it does allow for all these, like, weird invalid values to creep in. Because, in theory, you should never go beyond 2359. But even within the hours that are allowed, let's say, between 1000 and 1100, so between 10:00 and 11:00 a.m., a clock only goes up to 59 minutes. But our base 10 number system goes up to 99, so it's possible to have 1099, which is just an invalid time.

STEPHANIE: Right. And I imagine this isn't validated or anything like that. So it is possible to store some impossible time value in this database.

JOËL: I don't know for sure if the data is validated or not, but I'm not going to trust that it is. So I have to validate it on my end.

STEPHANIE: That's fair. One thing that is striking me is what time is zero?

JOËL: So zero in military time or just 24-hour clocks in general is midnight. So 0000, 4 zeros, is midnight. What gets interesting, though, is that because it's an integer, if you put the number, you know, 0001 into the database, it's just going to store it as 1. So I can't even say, oh, the first two digits are the hours, and the second two digits are the minutes. And I'm actually dealing with, I think, seconds and then some fractional part of seconds afterwards. But I can't say that because the number of digits I have is going to be inconsistent.

So, first, I need to zero pad. Well, I have to, like, turn it into a string, zero pad the numbers so it's eight characters long. And then, start slicing out pairs of numbers, converting them back into integers, validating them within a range of either 0 to 23 or 0 to 59, and then reconstructing a time object out of that.

STEPHANIE: That sounds quite painful.

JOËL: It's a journey for sure.

STEPHANIE: Do you have any idea why this is the case or why it was created like this originally?

JOËL: I'm not sure. I have a couple of theories. I've seen this kind of thing happen before. And I think it's a common way for developers who maybe haven't put a lot of thought into how time works to just sort of think, oh, the human representation. I need something to go in the database. On my digital clock, I have four digits, so why not put four digits in the database? Simple enough. And then don't always realize that there's all these edge cases to think about and that human representations aren't always the best way to store data.

STEPHANIE: I like how you just said that that, you know, we as humans have developed systems that are not quite, you know, the same as how a computer would. But what was interesting to me...something you said earlier about time being a fixed point. And that is different from time being a value, right? And so here in this situation, it sounds like we're storing time as a value, but really, it's more of the idea of, like, a point.

JOËL: Interesting. What is the difference for you between a point and a value?

STEPHANIE: I suppose a value to me...And I think we talked about this a little bit on a previous episode about value objects and also how we stored numbers, like phone numbers and credit card numbers and stuff like that. But a value, like, I might want to do math on. But I don't really want to do math on time. Or, specifically, if I have this idea of a specific point in time, like, that is fixed and not something that I could mutate and expect it to be the same thing that I was trying to express the first time around.

JOËL: Oh, that's interesting because I think when it comes to time and specifically points in time, I sometimes do want to do math on them. And so, specifically, I might want to say, what is the time that has elapsed between two points in time? Maybe I have a start time and an end time, and I want to say how much distance is there between the two? If you use this time system where you're storing it as an integer number where the digits have positional values, because there's all those gaps between, you know, 59 and 99 that are not valid, math breaks down. You've broken math by storing it that way.

So you can't get an accurate difference by doing math on that, as opposed to if you store it as a counter, which is what databases do under the hood, but you could do manually. If you just wanted to use an integer column, then you can do math because it's just a number of seconds since the beginning of the day. And you can subtract those from each other. And now you have these number of seconds between the two of them. And if you want them in minutes or hours, you divide by six here, 3600, and you get the correct response.

STEPHANIE: Yeah, that is really interesting because [chuckles] in this situation, you have the worst of both worlds, it seems like. [laughs]

JOËL: The one potential benefit is, I think, it's maybe more human-readable. Although, at that point, I would say if you're not doing math on it and you want something human-readable, you probably don't want an integer. You probably want a string. And maybe you even store it as, like, ISO 8601 time string in the database, or even just hour:minute:second split by a colon or whatever it is but just as a string. Now it's human-readable.

You can still sort by it if you go from largest to smallest increment in your format. You can't do math, but then you weren't doing math on it anyway. So that's probably a nice compromise solution. But, ideally, you'd use a native, you know, time of day column or a date-time or something like that.

STEPHANIE: For sure. Well, it sounds like something fun to contend with. [laughs]

JOËL: One thing that was brought to my attention that I'd never heard about before is that potentially a reason it's stored that way is because of an old data format called EDI—I think it's Electronic Data Interchange—that dates from ages ago, you know, the '60s or '70s, something like that. Before, we had a lot of standards for data; this is how...an emerging standard that came for moving data between systems. And it has a lot of, like, weird things with the way it's set up.

But if you're dealing with any sort of older data warehouses or older business systems, they will often exchange. And sometimes, you're going to store data in something that approximates this older EDI format. And, apparently, it has some weirdness around dates where it kind of does something like this.

So someone was suggesting, oh, well, if you're interacting with maybe an older, you know, a lot of, like, e-commerce platforms or banking systems, probably airline systems, the kind of things you'd expect to be written in, let's say, COBOL...

STEPHANIE: [laughs]

JOËL: Have a system that's kind of like this. So maybe that wouldn't be quite as surprising.

STEPHANIE: Yeah, that is really interesting. It just sounds like sometimes you're limited by the technology that you're interacting with. And I guess the one plus side is that, in your system, you can make the EDI work for you, hopefully. [laughs] Whereas perhaps if you are talking to some of those older technologies that don't know how else to convert date types and things like that, like, you just kind of have to work with what's available to you.

JOËL: Yeah. And that's got me realizing that a lot of these older, archaic systems are still online and very much a part of our software ecosystem and that there's a lot of value in learning some software history so that I'm able to recognize them and sort of work constructively with them when I have to interact with that kind of system.

STEPHANIE: Yeah, I really like that mindset.

JOËL: So, Stephanie, what's new in your world?

STEPHANIE: So, last week, we talked about writing reviews for ourselves and our peers. And one thing that happened in between the last episode and this one is Steph Viccari, former co-host of this podcast, who I've been working with really closely on this project of mine; she was writing a peer review for me. And one thing that she did that I really loved was she sent me a message and asked me a few questions about the direction of the review that I was wanting and what kind of feedback would be helpful for me.

And some of the things she asked were, you know, "Is there a skill that you're actively working on? Is there a skill you'd like to start working on?" And, like, what my goals are for the feedback. Like, how can she tailor this feedback to things that would help my progression and what I hope to achieve? And then my favorite question that she asked was, "What else should I know but didn't think to ask?" And I thought that was a really cool way of approaching.

You know, she's coming to this, like, wanting to be helpful, but then even still, like, there are things that she knows that I am kind of the expert on in my own career progression, and I really liked that. I think I'd mentioned last week that part of the feedback you want to be giving is, you know, something that will be helpful for that person, and centering them in it, instead of you is just a really awesome way to do that. So I was very appreciative that she asked me those questions.

JOËL: That's incredibly thoughtful. I really appreciate that she sent that out to you. What did you respond for the is there something else I should know but didn't know to ask?

STEPHANIE: Yeah. I mentioned that more and more, I'm realizing that I am not interested in management. And so what would be really helpful for me was to ground most of the feedback in terms of my, like, technical contributions. And also, that one thing that I'm thinking about a lot is how to be an individual contributor and still have an impact on team health and culture because that is something I care about. And so I wanted to share that with her because if there are things that she can identify in those aspects, that would be really awesome for me. And that can kind of help guide her away from a path that I'm not interested in.

JOËL: I think having that kind of self-awareness is really powerful for yourself. But then, when you can leverage that to get better reviews that will help you get even further down the path that you're hoping to go, and, wow, isn't that just, like, a virtuous cycle right there that's just building on itself?

STEPHANIE: Yeah, for sure. I think the other thing I wanted to share about what's new in my world that has been just a real boost to my mood is how long the days are right now because it's summer in North America. And yesterday was the summer solstice, and so we had the longest day of the year. The sun didn't set until 8:30 p.m. And I just took the opportunity to be outside. I took a swim in the lake, which was my first swim of the season, which was really special. And my friend had just a nice, little, like, backyard campfire hang out. And we got to roast some marshmallows and just be outside till the sunset. And that was really nice.

JOËL: When you say the lake, is that Lake Michigan?

STEPHANIE: Yes, I do mean Lake Michigan. [laughs] I forget that some people just don't have a giant lake next to them [laughs] that they refer to as the lake.

JOËL: It's practically an inland sea.

STEPHANIE: Yes, you can't see the other side of it. So, to me, it kind of feels like an ocean. And yesterday, when I was in the water, I also was thinking that I felt like I was just in a giant bathtub. [chuckles]

JOËL: So I'm in New England, and most of the bodies of water here are not called lakes. They're called ponds.

STEPHANIE: Really?

JOËL: No matter the size.

STEPHANIE: Oh.

JOËL: I guess lakes is reserved for things like what you have that are absolutely massive, and everything else is a pond.

STEPHANIE: That's so funny because I think of ponds as much smaller in scale, like a quaint, little pond. But that's a really fun piece of regional vocabulary.

So one interesting thing happened on my client project this week that I wanted to get your input on because I've definitely seen this problem before, and still, it continues to crop up. But I was working on a background job that we were passing a Boolean value into as one of the parameters that we would then, you know, use down the line in determining some logic.

And we, you know, made this change, and then we were surprised to find out that it continued to not work the way we expected. So we got some bug reports that we weren't getting into one of the branches of the conditional based on that Boolean value that we were passing in. And we learned, after a little bit of digging, that it turns out that those values are serialized because this job is actually saved in --

JOËL: Oh no.

STEPHANIE: [chuckles] Yeah. It inherits from the ActiveRecord, actually, and is saved in our database. And so, in that process, the Boolean value got serialized into a string and then did not get converted [chuckles] back into a Boolean. And so when we do that if variable check, it was always evaluating to true because strings are truthy in Ruby.

JOËL: Right. The string false is still truthy.

STEPHANIE: A string false is still truthy. And we ended up having to coerce it into a Boolean value to fix our little bug. But it was just one of those things that was really frustrating, you know when you feel really confident that you know what you're doing. You're just writing a conditional statement. And it turns out the language beguiled you. [laughs]

JOËL: I've run into similar bugs when I'm reading from environment variables because environment variables are always strings. But it's common that you'll be setting some kind of flag. So when you're setting the environment variable, you're setting something to true or false. But then, when you're reading it, you have to explicitly check if this environment variable double equals the string true, then do the thing. Because if you just check for the value, it will never be false.

STEPHANIE: Right. And I kind of hate seeing code like that. I don't know; something about it just rubs me the wrong way because it just seems so strange, I suppose.

JOËL: Is it just, like, those edge cases where you specifically have to do some kind of, like, double equals check on a value that feels like it should be a Boolean? Or do you kind of feel a bit weird about the concept of truthiness in general?

STEPHANIE: I think the concept of truthiness is very hard to grasp sometimes. And, you know, when you're talking about that edge case where we are setting...we're checking if the string is the string true. That means that everything else is false, right? So, in some ways, I think it's just really confusing because we've expanded the definition of what true and false mean to be anything.

JOËL: That's really interesting because now you have to pick. Are you checking against the string true, or are you negatively checking against the string false? And those are not equivalent because, like you said, now you're excluding every other string. So, is the string "Hello, World" put you in the false branch or the true branch?

STEPHANIE: Who's to say? [laughs] I think a similar conundrum also occurs when we use predicate matchers in our tests. I think this is a gripe that I've talked about a little bit with others when we're writing tests and especially if we're writing a predicate method, and then that's what we're testing, right? We kind of are expecting a true or false value.

And when our test expects something to be truthy rather than explicitly saying that we expect the return value to be true, that is sometimes a bit confusing to me as well because someone could theoretically change this method and just have it return "Hello, World," like you said, as a string, like, anything else. And that would still pass the test.

JOËL: And it might even pass your code in most places.

STEPHANIE: Right. And I suppose that's okay. Is it okay? I don't know. I'm not sure where I land on this.

JOËL: I used to be a kind of hardcore Boolean person.

STEPHANIE: [laughs] That's a sentence no one has ever [laughs] said.

JOËL: I like my explicit trues and falses. I don't like the ambiguity of saying, like, oh, if person do a thing, it's, like, oh, what is person here? Is this a nil check? Is it explicitly false? Do you just want to know that this person is non-empty? Well, what exactly are you checking? So I like the explicitness of saying, oh, if person dot present, or if person dot empty, or if person dot nil.

And I think maybe spending some time in some more strongly typed languages has also kind of pushed me a little bit in that direction, where it's nice to have something that is explicitly either just true or just false. And then you completely eliminate that problem of, like, oh, but what if it's neither true nor false, then what do we do for that branch there? And the answer is your compiler will reject that program or say, "You've written a bad program." And you never reach that point where there's a bug.

I've slowly been softening my stance. A fellow thoughtbot colleague has written an article why there is no such thing as a Boolean in Ruby. Everything is just shades of gray and truthiness and falsiness. But from the perspective of a program, there is no such thing as a Boolean. And that really opened my eyes to a different perspective. I don't know that I fully agree, but I'm kind of begrudgingly acknowledging that Mike makes a good point.

STEPHANIE: Yes, I read the blog post that he wrote about this exact problem. And I think it's called "Booleans Don't Exist in Ruby." And I think I similarly, like, came away with, like, yeah, I think I get it if I just suspend my disbelief, you know, hard enough. [laughs] But what you were saying about, like, liking the explicitness, right? And liking the lack of ambiguity, right? Because when you start to believe that Booleans don't exist, I think that really messes with your [laughs] head a little bit.

And one takeaway that I got from that blog post, kind of like we mentioned earlier, is that there is such thing as false, and then everything else is true. And I guess that's kind of how Ruby operates.

JOËL: Sort of, because then you have the problem of nil, which is also falsy.

STEPHANIE: That's true, but nil is nothing. [laughs]

JOËL: That's one of the classic problems as well when you're trying to do a nil check, or maybe some memoization, or maybe even, say, cache this value, or store this value, or initialize this value if it's not set. And assuming that doing nil is falsy, so you'll do some kind of, like, or equals, or just some kind of expression with an or in it thinking, oh, do this extra work if it's nil because then it will trigger the branch. But that all breaks down if potential for your value to be false because false and nil get treated the same in conditional code.

STEPHANIE: Right. I think this could be a whole separate conversation about nil and the idea of nothingness. But I do think that, as Ruby developers, at least in the Ruby world, based on what I've seen, is that we lean on nil in ways that we maybe shouldn't. And we end up having to be very defensive about this idea of nil being falsy. But that's because we aren't necessarily thinking as hard about our return values and what our arguments are that; it ends up causing problems in evaluating truthiness when we're having to check those objects that could be nil.

JOËL: In terms of the way we communicate with the readers of our code, and, as a reader, I generally assume that a Ruby method that ends with a question mark will return a true Boolean, either true or false. Is that generally your expectation as well?

STEPHANIE: I want to say yes, but I've clearly experienced enough times where that's not the case that, you know, it's like, my ideal world and then reality [laughs] and having to figure out how to hold both of those things.

JOËL: It's one of those things that's mostly true.

STEPHANIE: I want to believe it because predicate methods and, like, the Ruby Standard Library mostly return Boolean values, at least to my knowledge. And if we all kind of followed that [laughs] pattern, then it would be clear. But I think there's a part of me that these days mostly believes it to be true that I will be getting a Boolean value (And, wow, even as I say this, I realize how confusing [laughs] this is starting to sound.) and that until I'm not, right? Until I'm surprised at some point.

JOËL: I think there's two things I expect of predicate methods in Ruby. One is that they will return, like, a hard Boolean, either true or false. The second is that they are purely query methods; they don't do side effects. Neither of those are consistent across the ecosystem.

And a classic example of violating that second guideline I have in my mind is the valid question mark method from Rails. And this really surprised me the first time I tripped into this because when you call that on an object, it doesn't just tell you whether or not the object is valid. It actually mutates the underlying object by populating the error messages' hash. So if you have an invalid object and you examine its error messages' hash, it will be empty until you call the valid question mark method.

So sometimes, you don't even care about the return value. You're just calling valid to mutate the object so that you can access the underlying hash, which is that's weird code when you call a predicate method but then totally ignore the output.

STEPHANIE: Yeah, that is strange because I have definitely seen it where we are calling the valid method to validate, and then we end up using the error messages that are set on that object later. I think that's tough because, in some ways, you do care about whether the object is valid or not. But then also, the error messages are helpful usually and when you're trying to use that method. The point is to validate it so that you can hopefully, like, tell the user or, like, the consumer of your system, like, what's wrong in validation. But it is almost, like, two separate things.

JOËL: It is. And sometimes, it's really hard to split those two apart. So I'm not throwing shade at the Rails dev team here. Some of these design decisions are legitimately difficult to make. And what's most useful for the most people the most time is often a compromise. I think you brought up the idea of separating those two things. And I think there's a general principle here called command-query separation. That's, like, the fancy way of talking about what you were saying.

STEPHANIE: One thing that I was just thinking about kind of when we initially picked off this conversation was the idea of how things outside the Ruby ecosystem or the Ruby world interact with what we're returning in terms of Boolean values. And so when I mentioned the object being serialized because of, you know, our database and, like, background job system, that's an entity that's figuring out what to do with the things that we are returning from Ruby.

And similarly, when you're talking about environment variables, it's like, our computer system talking to now our language and those things being a bit different. Because when we, like, suspend our disbelief about what is truthy or falsy in Ruby, at least we're doing it in, like, the world of Ruby. And as soon as we have to interact with something else, like, maybe that's when things can get a little hairy because there's different ideas about truthiness there. And so I'm kind of also thinking about what we return in APIs and maybe, like, that being an area where some explicitness is more required.

JOËL: Whenever I'm consuming third-party data, I'm a big fan of having some kind of transformation or parse step. This is inspired in part by the "Parse, Don't Validate" article, which I'll link in the show notes. So, if I'm reading data from a third-party API and I want it to be a Boolean, then maybe I should do the transformation myself. So maybe I check literally, is it the string true or the string false, and anything else gets rejected?

Maybe I have...and maybe I'm a little bit more permissive, where I also accept capital T or capital F, and I have, like, some rules for transforming that. But the important thing is I have an explicit conversion step and reject any bad output. And so for something like an environment variable, maybe that would look like looking for true or false and raising if anything else is there. So that we try to boot the app, and it immediately crashes because, hey, we've got some, like, undefined, like, bad configuration that we're trying to load the app with. Don't even try to keep running. Hard crash immediately. Fix it, and then come back.

STEPHANIE: Yeah, I like that a lot because the way we ended up fixing this issue with the background job that I mentioned was just coercing our string value into a Ruby Boolean in the job that we were then, like, running the conditional in. But really, what we should have done is have fixed that at a higher level and where we parse and deserialize, like, the values we're getting from the job to prevent this kind of in the future because right now, someone can do this again, and that's a real bummer.

JOËL: I always love those deeper conversations that happen after you've had a bug that are like, how do we prevent this from happening again? Because sometimes that's where you have the deepest learnings or the most interesting insights or, you know, ideas for Bike Shed episodes.

I'm really curious to contrast JavaScript's approach to truthiness to Ruby's because even though they both use the same idea, they kind of go about it differently.

STEPHANIE: Tell me more.

JOËL: So, in Ruby, an empty array and an empty string are truthy. JavaScript decided that empty things are falsy. And I forget...there's a whole table that shows the things that are truthy and falsy in JavaScript. I want to say zero is falsy in JavaScript but don't quote me on that, which can also lead to some interesting edge cases you have to think about.

STEPHANIE: Okay, yes. This is coming back to me now. I think depending on what, you know, ecosystem or language or world I'm in, I have to just only be able to think about what is true in this world [laughs] and then do that context switching when I am working in something else. But yeah, that is a really interesting idea. Someone decided [laughs] that this was their idea of true or false.

JOËL: I'm curious if you have a preference for sort of JavaScript's approach to falsiness where a lot more types of values are falsy versus Ruby, which said pretty much only nil and false are falsy. Everything else is truthy.

STEPHANIE: Hmm, that is an interesting question.

JOËL: Because in Ruby then or, I guess, in Rails, we end up with the present predicate method that is specifically checking for not only nil and false but also for empty array, empty string, those kinds of things. So, if you find yourself writing a lot of present matchers in your code, you're kind of leaning on something that's closer to JavaScript's definition of falsiness than Ruby's. But maybe you're making it more explicit.

STEPHANIE: Right. In JavaScript, I see a lot of double bangs in lieu of those predicate methods. But I suppose by nature of having to write those predicate methods in Ruby, we're, like, really wanting something else, I think. And maybe...I guess it is just a question of explicitness like you're saying, and which I prefer. Is it that I need to be explicit to convey the idea that I want, or is it nice that the language has just been encoded that way for me?

JOËL: Or maybe when you write conditionals, if you find yourself doing a lot of presence checks, do you find that you typically are trying to branch on if not null, not false, not empty more frequently than just if not null, not false? Because that's kind of the difference between Ruby's model and JavaScript's model.

STEPHANIE: Hmm, the way you posed that question is interesting because it makes me think that sometimes it's quite defensive because we have to check for all these possible return values. We are unsure of what we are getting back. And so this is kind of, like, a catch-all for things that we aren't really sure about.

JOËL: Yeah, I mean, that's the fun of dynamic programming languages. You never know exactly what you're going to get as long as things respond to certain methods. You really lean into the duck typing. And I think that's Mike's argument in his article that "Booleans Don't Exist" in that as long as something is responding to methods that you care about, it doesn't matter if you're dealing with a true Boolean or some kind of other value.

STEPHANIE: Right. So I suppose the ideas of truthiness then are a little bit more dependent on how people are using the language though it seems like a chicken-and-egg situation to me. [laughs]

JOËL: It is really interesting to me in terms of maybe thinking about use cases in my own code if I'm having to...if I'm writing code that leans on truthiness where I can say just, you know if user. But then knowing that, oh, that doesn't account for, like, an empty value. Do I then also need to add an extra check for emptiness? And maybe if I'm in a Rails project, I would reach for that present matcher where I wouldn't have to do that in JavaScript because I can just say, if user, and that already automatically checks for presence.

So I'm kind of wondering now in my mind, like, which default would fit my use cases more? Or, if I go back to an older version of myself, I will say I don't want any of these defaults. They're all too ambiguous. I'm going to put explicitly if user dot nil question mark, if that's the thing that I'm checking for, or if user dot empty question mark because I want my reader to know what condition I'm checking.

STEPHANIE: Yeah, that is interesting, this idea of, like, which mode do you find yourself needing to use more and if that is accommodated for you because that's just the more common, like, use case or problem.

I think that's something that I will be thinking about the next time I write a conditional [laughs] because, like I was saying earlier, I think I end up just leaning on what someone else has decided for me in terms of truthiness and not so much how I would like it to work for me.

JOËL: And sometimes we don't want to fight the language too much, you know if I'm writing Elm, that everything is hard Booleans. And I know I'm never going to get a nil in a place where I'd expect true or false because the compiler would prevent that from happening. I know that I'm not going to get an empty value, potentially.

There's ways you can do things with a type system where you can explicitly say no empty values are even allowed at this point. And if you do allow them, then the type system will say, "Hey, you forgot to check for the empty case. Bad program. I'm rejecting that." And then you have to write that explicit branch for, oh, if empty versus if present. So I really appreciate that style of programming.

But then, when you're in a language like Ruby where you're not dealing with explicit types on purpose, how do you shift that mindset so that you don't need to know the type of the value that you're dealing with? You only want to say, hey, in this context, here's the minimal interface that I want it to conform to. And maybe it's just the truthy or falsiness interface, and everything beyond that is not relevant.

STEPHANIE: I think it's kind of wild to me that this idea of a binary that theoretically seems very clear turns out is actually quite confusing, ambiguous, philosophical, even. [chuckles]

JOËL: Yeah. It's definitely...you can get into some deep, philosophical questions there, language design as well.

One aspect, though, that I'm really curious about your thoughts is bringing new people in who are learning a language. It's really common for people who are learning a language for the first time, learning to code for the first time to write code that you and I would immediately know, like, that's not going to work. You can't add a Boolean and a number. You're just learning to code. You've never done that before. You don't know. And then how the language reacts to that kind of thing can help guide that experience.

So, do you think that truthiness maybe makes things more confusing for newcomers? Or, maybe on the other side, it helps to smooth that learning curve because you don't have to be like, oh, wait, I have a user here. I can't put that in a condition because that's not a strict true or false. I'm going to coerce it, or I've got to find a predicate method or something. You can just be like, no, put it in. The interpreter will figure it out for you.

STEPHANIE: Wow. That's a great question. I'm trying to put myself in the beginner's mindset a little bit and think about what it's like to just try something and the magic of it working. Because, like you said, the interpreter does it for you, or whatever, and something happens, and you're like, wow, like, that was really cool. And I didn't have to know all of the ins and outs of the types of things I was working with.

That can be really helpful in just getting them, like, started and getting them just, like, on the ground writing code. And having that feeling of satisfaction that, like, that they didn't have to, you know, have to learn all these things that can be really scary to make their program work.

But I do think it also kind of bites them later once they really realize [laughs] what is going on and the minute that they get that, like, unexpected behavior, right? Like, that becomes a time when you do have to figure out what might be going on under the hood. So two sides of the same coin.

JOËL: What you're saying there about, like, maybe smoothing that initial curve but then it biting them later got me thinking. You know how we have the concept of technical debt where we write code in a way that's maybe not quite as clean today so we can move faster but that then later on we have to pay it back? And I almost wonder if what we have here is almost like a pedagogical debt where it's going to cost us a month from now, but today it helps us move faster and actually kind of get that momentum going.

STEPHANIE: Pedagogical debt. I like that. I think you've coined a new term. Because I really relate to that where you learn just enough to do the thing now. But, you know, it's probably not, like, the right way or, like, the most informed—I think most informed is probably how I would best describe it—way of doing it. And later, you, yeah, just have to invest a little more into it. And I think that's okay.

I think sometimes I do tend to, like, beat myself up over something down the line when I have to deal with some piece of less-than-ideal code that I'd written earlier. Like, I think that, oh, I could have avoided this if only I knew. But the whole point is that I didn't know. [laughs] And, like, that's okay, like, maybe I didn't need to know at the time.

JOËL: Yeah, and code that's never shipped is of zero value. So having something that you could ship is better than having something perfect that you didn't ship.

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: Byeeeeeeee!!!!!!

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.

Support The Bike Shed

Förekommer på
00:00 -00:00