text stringlengths 0 2.35k |
|---|
**Kris Brandow:** I think there was also a question about how we might start quantifying what this debt looks like for the business. And I think kind of using their own terms would be helpful here. Businesses tend to have really good -- or at least the ones that survive for a long time, tend to have really good risk ma... |
Once we start talking in less of these absolute terms, I think it becomes easier for us to communicate with businesspeople and give them something quantifiable. It's like, "Okay, if we don't do this work, there's an 80% chance that we're gonna run into a problem", versus like "If we don't do this work, there's a 20% ch... |
\[39:52\] So we have to say "Okay, we have an 80% probability of this thing happening. Okay, did it actually happen, and how many times did it actually wind up happening?" Just going back and tracking all of that and tracing all of that, so then every time the business asks, we can be like "Here's the pages and pages a... |
But I think, to your point of that type of engineer that we need to develop, I think one of the types of engineers we do need to develop are people that are kind of focused on going into codebases and doing this type of analysis, and figuring out how to prioritize that debt that we have and come up with debt consolidat... |
**Jon Sabados:** That is definitely an interesting thought, and like you mentioned, that is almost like a new type of engineer, a new type of position, because coming up with all those percentages and whatnot - that's like an FTE almost. It's not something that you can just kind of do on the side, while you're doing yo... |
**Kris Brandow:** Yeah, absolutely. It's a thing you have to be dedicated to doing, and actually want to push your organization forward with doing in the future. And it's like a whole process and a whole framework that you have to develop as an engineering team. We're lacking this in most of our engineering organizatio... |
**Jon Sabados:** Yeah, and I wonder if that also calls up -- there's oftentimes so much ceremony involved; at least my experience has been a lot of engineers are like "Well, this is kind of killing productive time, because if we spend a bunch of time pointing things, and we do the work based on scrums, but no one's hol... |
**Kris Brandow:** Yeah. I think it's the kind of hole we've dug for ourselves in, building a system where we are supposed to be getting these benefits, but then not actually the back-half of that work. Not actually figuring out, "Is this form of agile we're doing, are sprints working for us? Are storypoints working for... |
**Angelica Hill:** I feel a whole Go Time episode coming on on this exact topic. So I'm interested to hear a little bit more about -- we've talked about getting to that explosion point where it's falling apart, having bugs every day, making on-call engineers' life hell; that is the point at which you're like "We need t... |
**Jon Sabados:** I think one thing if you wanna keep from code becoming legacy, I think that testing is a helpful tool for it. It's by no means a solve it, but if you can define your actual problems that you're trying to solve, and the right tests that exercise those problems, that will also help ensure that you've got... |
**Jeff Hernandez:** \[43:58\] It sounds like you're advocating for test-driven development... Is that right? |
**Jon Sabados:** I just might be! |
**Jeff Hernandez:** \[laughs\] |
**Jon Sabados:** It's the cure-all, and there's oftentimes a lot of places where test-driven development doesn't work, especially like lab settings or where you don't know what you're going to do, but it is a tool that helps on more than just like ensuring that your code functions correctly. |
**Misha Avrekh:** In my experience, you have to invest in your infrastructure and your monitoring probably before you invest in actually rewriting your code... In the sense that for example if your deployment pipeline takes hours for your changes to get deployed, you need to rewrite the deployment pipeline first, right... |
**Dominic St-Pierre:** Yeah, I like that. Going small piece by piece, and just being in a better way each week probably than you were last week; it feels like a great way to do that, and also probably less scareful and dangerous, if I can say that... Because when you think about that rewriting as an entire software, it... |
**Kris Brandow:** I think I would say writing documentation, or just comments, really. I think testing is definitely helpful, but I think if I were to be given a codebase that has really good tests, but no comments and documentation, or a codebase that has documentation and comments, but no tests, I would definitely ta... |
But when I only have tests and they don't have good documentation and comments, then it's like, I have to assume that the tests were also written to do the thing that was intended to be done, or kind of try and derive from those tests what's supposed to be handled here. |
But I think just going through and being like, okay, here's a function with this name. Here's a type with this name. Here's a method with this name. Does this do the thing I think it does, and then writing that down and making sure everybody has an understanding of that - a real great way to actually a) start exploring... |
**Jon Sabados:** Clarity of intent is a good one, because that's also like comments, documentation, and even just the coding style as well can go a long way towards that... Because naming is one of those things that's really hard, and people tend to just eventually give up on, or get frustrated and be like "Ah, whateve... |
**Dominic St-Pierre:** I think that's another strength of Go, in my opinion, due to the packages and whatnot, compared to other frameworks and languages that I've worked with in the past. So you kind of already have a small sense of what something is going to do in terms of business case and whatnot, because they are p... |
**Jeff Hernandez:** I'm wondering if people have any examples... Because everything so far sounds like having good engineering standards and protocols for your team and organization is a big help in preventing legacy code, I'm curious if others have examples of maybe not just good documentation or good tests and stuff ... |
**Kris Brandow:** \[47:59\] I've definitely found that, at least for me -- well, the outcome of this is usually writing better documentation, or tests, or whatever... But I think just like taking some extra time to stop and really think about what you're trying to do and what you're trying to implement. It doesn't have... |
And I think also just many iterations of things. So if you write something and you budget your time and plan on being able to write it 3, 4, 5 times, that gives you the space to like write it once, maybe quick and dirty and look at it and say "But is this the way that it should be written?" and kind of get that feeling... |
I think a lot of the times we wind up with legacy code, or the bad legacy code, because we didn't go through enough iterations of thinking about something... And I think it can be very difficult to sit down and do that, because we do have deadlines we're trying to hit, but I think that's one of the things that has defi... |
I think a lot of times, once again, we ignore that thread, because we have things to do, we have features to develop, and we have something that works, so why are we gonna follow this thread; but I think really just following that thread at the end of the day has definitely helped me write codebases that are much more ... |
**Dominic St-Pierre:** Do you think that pair programming could also do this effect? |
**Kris Brandow:** I think it can, if your environment is one that is, I'd say, opportune for pair programming. If you're actually in person, you can actually sit at the same desk, you can actually work through things together at the same time, I think then yeah, pair programming can be very helpful. I think also just g... |
So yes, I definitely think pair programming can be excellent for this. I think whiteboard sessions, if you can be in a space where you can do whiteboarding, is great. But I think also just - maybe not doing the pair programming part of programming, but just hopping on a call and talking through an idea with people, may... |
**Angelica Hill:** So we've talked a little bit about kind of you have this legacy thing, and how do you make it so that you don't get to that final explosion point... But I'm interested to talk a little bit about when/if you get that buy-in to rewrite a legacy codebase, or to change up a system and business, for whate... |
**Kris Brandow:** \[52:08\] I would say from the jump understand what went wrong with your old system. I think people run for microservices or for Kafka, for some new, flashy tech, at the end of the day; they're like "Oh, the problem was we weren't using microservices", or something. But it's like, sit down and actuall... |
A thing that I have done, that has hilariously sometimes wound up with me not actually abandoning old codebases when I really wanted to, because the problems I wanted to fix I could just fix in that old codebase, so why make a whole new one, with all new problems... But I think that's definitely where I tend to start, ... |
**Angelica Hill:** Jon, what do you think? I know this is something that I'm hoping you're thinking about... |
**Jon Sabados:** I don't know if I've got too much on this one... |
**Angelica Hill:** Also, off the record, for those who didn't see my cheeky smirk, Jon and Jeff in fact are both on my team... So when I make any cheeky comments in their direction, it's slightly more layered than other people who don't have that context may know. And I am their technical product manager, so that's als... |
**Kris Brandow:** Undercover businessperson. |
**Angelica Hill:** I'm undercover. Secret gopher... Well, over gopher, secret product manager. \[laughs\] Jeff, what are some questions that you would hope -- if I was to go to you, Jeff, and be like, "Hey, Jeff, I'm thinking we should completely get rid of our current platform and rewrite it." What would be things tha... |
**Jeff Hernandez:** Yeah, I'm definitely following what Kris is saying. You definitely have to understand why you're trying to rewrite everything. You don't wanna rewrite just for the sake of "Oh, we have lots of issues with this codebase. Maybe we should just declare bankruptcy and start something new." |
It's definitely something I myself have been interested in. For our team specifically, we're in the middle of rewriting our two platforms, and getting some historical context as to why we went that route. It would be interesting. But I don't wanna trigger Jon's PTSD too much, because it looks like he's going through so... |
**Jon Sabados:** Yeah, and I guess rewrites are tough... I feel like everybody at some point in their career needs to go through the "Let's just do the rewrite because of it, and feel the pain, and watch it fall flat on its face, and waste like six months of our lives..." |
**Kris Brandow:** Six months if you're lucky. |
**Jon Sabados:** Yeah. I've seen entire years go down the drain. So having an understanding of the risk involved is huge. And you know, there are scenarios that do warn it, but they're the exception rather than the rule. |
**Angelica Hill:** So am I right in assuming that if your product partner or the business came to you - I'm speaking to all of you here - and said "Right, Kris, Misha, Dominic, I know you've been working with this 20+ year (in Dominic's case) infrastructure. We're gonna give you five years maximum to rewrite completely... |
**Dominic St-Pierre:** I would only rewrite a software that works if the technology is not going to exist sooner. Take an old system like Cobol. In Quebec there's a couple of companies that are still built on top of AS/400 and Cobol and whatnot. I mean, there's no manpower anymore for that, so what do you do? You don't... |
**Kris Brandow:** \[56:22\] I'm in agreement there. I think when it comes to greenfield projects, or when it comes to rewriting legacy projects, just don't. I think that's kind of the hard and fast rule most of the time. Because sure, usually we get to the point we want to rewrite something when there's 20% to 30%, may... |
When you do a rewrite, you have to account for all those things that it does correctly, and then replicate them, and then also fix all those bugs. So it's always going to take way, way, way longer to rewrite the thing than it is to fix the problems in the thing that exists. So I think that's the thing that we have to a... |
**Jon Sabados:** Yeah, there's gotta be some fundamental problem that requires a complete greenfield rewrite. It can't just be bang where everyone is gonna be like -- And also, it shouldn't come from product. It should be tech being like "There's this technical roadblock that means we just cannot do this anymore." Beca... |
**Kris Brandow:** If you can't come up with a way of conceivably, actually retiring completely the thing that you're rewriting at the beginning of your rewrite, then you shouldn't do it. Because you're not going to be able to retire it if you don't already have a plan for how to do it. I think that's just like fact at ... |
**Dominic St-Pierre:** \[01:00:20.26\] You know how it is, even if everyone is agreeing to not touch the old thing anymore, they will always have a new feature, there will always be bug fixes that you will need to replicate on the new system. It's extremely difficult. |
**Kris Brandow:** Yeah, don't assume that you will not have to add something new to the old system until the old system is completely off. So if your new thing is "I'm gonna go build something new, and it's gonna take me 2,5 years to build it", assume that you have to keep adding stuff to the old thing until that 2,5 y... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.