text stringlengths 0 2.35k |
|---|
**Egon Elbre:** Yeah. But there's one funny thing there... Let's say there's an innovation in the technology, right? When React came up, the idea of maintainability or the effort to maintain got lower, which means that your technical debt goes high up if there's an innovation. So... |
**Roger Peppe:** That's interesting. And does that mean that when you start a project and it's not by any means finished, that that counts as technical debt, because you just haven't finished that project that you've just started? |
**Roger Peppe:** Yeah, I guess... Let's say you have a security debt; you haven't done a proper security audit, right? There are other aspects too you can measure. And if you look at maintainability only, then an unfinished project isn't necessarily in debt because of it being unfinished. So that ideal state in terms o... |
**Mat Ryer:** Yeah. That is interesting. So in that case -- because this is the other thing... If you have what you perceive to be a technical debt, and you acknowledge it, and you decide "For priority reasons, we're going to leave it", and then later you find out "We actually didn't need that. We never need it", then ... |
Yeah... And so I think one of the other principles is designing things so that you can change them. Designing things for that flexibility in the future. I think it's quite an interesting idea. I think I saw it on Twitter, somebody said, like, architects - architects are there, and it sounds great, because they're desig... |
**Roger Peppe:** That said, I think it really helps to do a lot of thinking through. You can actually design stuff; it might not be set in stone, but I do think that you can eliminate a lot of bad code paths or bad futures by just thinking them through to start with, and thinking, "Well, if we went that direction, wher... |
I think a lot of people are like "We shouldn't do any design upfront." I really don't agree with that approach at all. If you look at Go, for example - they designed a lot. They wrote the spec before they did the implementation. |
**Mat Ryer:** Yeah. |
**Egon Elbre:** I think there was one quote that when you're designing things, add flexibility into the places where you are most uncertain. |
**Roger Peppe:** Yep, that's a really good point. |
**Mat Ryer:** Yeah, that's true. |
**Roger Peppe:** That sort of leads through to one of the things where I think is sort of key to avoiding code bloat, is just designing the right abstractions. If you get the abstractions right up front... There's this feeling when you're wrestling with the wrong abstraction, you just feel the code is piling up. You're... |
**Egon Elbre:** \[41:52\] I remember a talk about designing things for deletability, so that it's easy to delete features, rather than to extend... Because if it's easy to delete, then it's probably easy to replace. |
**Roger Peppe:** Yep. |
**Mat Ryer:** Yeah, I like that point, actually, on the abstraction thing... I talk a lot about kind of avoid doing the abstractions too early. Do a few examples first, and see. It'll depend on the project or the problem. But actually, the real value is in these abstractions. That is really probably one of the most val... |
So that's the other thing, it is very important, and practicing designing abstractions, and honestly, getting it wrong; practicing, and making those mistakes, and living with the pain, and then you learn from that. And I think there is an element of takes -- it takes a lot of experience. I don't think any of these gene... |
**Roger Peppe:** If anyone wants a good exercise for practicing abstractions - people I mentor, I usually give these exercises. You need to write temperature conversion code. But now you need to write it in 20 different places, and they have to be conceptually different design-wise, and the different properties that th... |
**Mat Ryer:** That'd be a good talk... |
**Roger Peppe:** That sounds like an interesting exercise, yeah. |
**Mat Ryer:** Yeah. I'd like to see you do that in a talk. |
**Roger Peppe:** Then he would ruin all the future things for his mentors to learn... |
**Egon Elbre:** Oh no, I can pick any different problem... |
**Mat Ryer:** Yeah, the temperature thing reminds me... In my school, I was the first year to do IT as a GCSE, but the school just wasn't ready for it. No one really had computers. They were teaching how to do spreadsheets, and word processing, and things. But I was into computers from a very young age, so I really lov... |
**Roger Peppe:** Input and output. |
**Mat Ryer:** You can only pick one, Roger. Otherwise you get no points. |
**Roger Peppe:** Oh, you can only play one? Oh, okay. |
**Mat Ryer:** Oh my, feel my pain. |
**Roger Peppe:** Oh, no... |
**Mat Ryer:** The worst was hard drive. Hard drive - it's like, how are you getting things on and off if it's not also input and output? I left those notes in the side, on the exam, of course... |
**Roger Peppe:** Of course you did. |
**Mat Ryer:** Yeah. Pedantic. But the last question was -- it just said, "Design or explain how you would regulate the temperature in a swimming pool." And I wrote the BASIC code; because I used to write BASIC at home. So I wrote this little BASIC program that did that... And I don't know, it was just like -- we didn't... |
**Break:** \[45:15\] |
**Mat Ryer:** Egon, you made a point earlier about features... Because I think that is another way to avoid bloat. And a feature, when you imagine it, say, in a website, and it shows up - it just seems like a logical thing to have. But sometimes that feature adds quite a lot of complexity to the system. |
**Egon Elbre:** Yeah. |
**Mat Ryer:** And considering the actual cost, the engineering cost, the maintenance cost really, of features, I think it's something that gets overlooked a lot. Where I am at Grafana, because all the leadership, we're engineers, that is understood from the beginning. It's implied we are always thinking about that. I'v... |
**Egon Elbre:** I think one of the issues with features is that they are really hard to remove afterwards. |
**Mat Ryer:** Yeah. Because people rely on them, of course. |
**Roger Peppe:** Yeah. You basically can't remove a feature when you've added it. The other thing is that interaction between features - that also leads to code bloat, because these things are maybe non-orthogonal, you're changing this other thing in the codebase, and that interacts with all these other features, which... |
**Mat Ryer:** Yes. That is also a thing I think people should always ask themselves when you're considering how to solve a problem. This is why I always like it whenever use cases come in, when they focus on the problem. It's very easy for people to write a ticket to say, "Oh, we need a button here that does this. This... |
**Roger Peppe:** Yeah. A user story. |
**Mat Ryer:** Yeah. Really, we want to know what's the problem that they're trying to solve, and then take a step back and see, like you say... And it's like, I can think of so many examples of places where we've had that thought, we found the general one, and then it's just paid dividends again and again and again dow... |
**Roger Peppe:** Absolutely. |
**Mat Ryer:** So yeah, I think that's great. |
**Roger Peppe:** That whole take a step back thing is actually key throughout all of software engineering, I think. Don't just focus directly on what you're trying to solve; think through the code and in the wider situation. When you're reviewing code -you know, this code is addressing this particular problem, but mayb... |
**Mat Ryer:** Well, it's that time... It's that time that -- we always have a time on Go Time, and it's this time... It's unpopular opinions time. It's time also for the theme tune. |
**Jingle:** \[50:26\] to \[50:46\] |
**Mat Ryer:** Roger, would you record a fiddle accompaniment to that, so that we can play it next time? |
**Roger Peppe:** Alright. Done. |
**Mat Ryer:** Oh, that's gonna be great... Thank you. \[laughs\] And we'll get your TinyGo music machine as well, Egon, which we already have a recording of... Okay, so who's got an unpopular opinion for us today? |
**Egon Elbre:** I think I already said it, but you should review every line - it's either direct or indirect of your dependencies - with similar standards as your own codebase. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.