Robby Russell on Oh My Zsh, Developer Experience, and Open Source

1 - > Robby: I never set out to think that this is gonna be a popular 2 - > widely used tool. Like, I literally wanted, like, eight of 3 - > my coworkers to have this on their computer so that I could 4 - > be lazy and not have to remember all these complex Git commands. 5 - > And I just shared it publicly, and it kind of spread. 6 - > Jack: Robby Russell created OmoZ Shell just to share his configs 7 - > with his colleagues.

Now it's one of the most popular open 8 - > source projects in the world. In this episode, we talk about the 9 - > origins of OmoZ Shell, developer experience, and the future of 10 - > developer agencies. Enjoy the episode. So I remember, like, 11 - > when we spoke before, you were kind of adamant that it was 12 - > never meant to be a popular project, and it was just for 13 - > your colleagues initially.

14 - > Robby: Exactly. So the the kinda like the initial impetus of the 15 - > project was I had been using Z Shell for a few years on my 16 - > computer, and this was back. So the project started I believe it 17 - > was 2009, like maybe the, like August I think is maybe the 18 - > first commit to the repository, and a repository was created. 19 - > But I had been working Z Shell on my in my CLI for maybe one 20 - > and a half, two years or so, and I would had been introduced to 21 - > it from some other developers in over IRC channels and the Ruby 22 - > on Rails community that I had been part of since, like, 23 - > beginning 2005, And it had a Git was still kind of a new thing at 24 - > the time.

It was maybe like a year. 25 - > I've been using Git maybe for about a year, year. I think 26 - > GitHub was less than a year old at that point as well. And I had 27 - > gotten, I built up all these like aliases and shortcuts in 28 - > this and kinda clobbered together this mess of a 29 - > configuration file from copy and pasting for my friends, from 30 - > things that I don't know if you remember things like pasty.

org, 31 - > think it was the website. There's like a gist type sites 32 - > where people would put little like, they'd throw up their 33 - > config files, and like, here's how I did this, and then so we 34 - > were kinda copying and pasting and trading, like Pokemon cards 35 - > or something, like, had collect them all and like so I had this 36 - > very complicated several 100 line configuration file for my Z 37 - > shell configuration. And it gave me all these cool things like 38 - > colors in my prompt and like the it would show like the git 39 - > branch that I was using because local branches were kind of like 40 - > a new novel thing for software developers because prior to 41 - > that, we'd be using Subversion or CVS, and you you didn't 42 - > really think about it.

43 - > Like, why would I need a local branch versus like a branch on, 44 - > you know, your your source code management tool, like like that 45 - > maybe your team can collaborate on, but why would I need a local 46 - > branch? But so just to kinda get used to getting familiar with 47 - > git, I had all this stuff to kinda make that experience a 48 - > little bit better. And then a whole bunch of aliases and 49 - > shortcuts, but then I'd be pair programming with my coworkers on 50 - > their computer, and they'd maybe be still using bash because that 51 - > was the default shell on their computer, and I couldn't 52 - > remember a lot of the shortcuts.

You know, like I was like, the 53 - > muscle memory wasn't there. So I'm like, just know how to type 54 - > all these like three little three characters and it 55 - > generates this git command that I couldn't remember all because 56 - > git has a lot of like different methods and shortcuts. 57 - > Know there were a lot of there's a lot of git offers a lot of 58 - > functionality and you couldn't remember, like, here's like the 59 - > five or six little shortcuts I'd use and I couldn't remember 60 - > them.

So we were constantly having to, like, look them up. 61 - > Like, what's the command for that? And and so I was like, you 62 - > know, if you just copy my Z shell configuration file and 63 - > switch over, then you could take advantage of all my cuts as 64 - > well. Some of the team, my team my coworkers would would would 65 - > just copy and paste it and then switch over to using Z Shell as 66 - > well.

67 - > Great. And then I could kinda be lazy and can kinda keep doing my 68 - > thing if I was the one driving or teach them how to use it, but 69 - > a couple of my coworkers were very reluctant to do that 70 - > because they didn't understand what the config file did. And 71 - > like, well, wanted to understand how it all worked and and they 72 - > wanted me to walk him through it. And so I was like, well, if 73 - > I'm being honest, I've been copying and pasting and making a 74 - > few tweaks here and there, I didn't fully understand it 75 - > myself, so I was like, well, if I can't really walk you through 76 - > it myself, so how about I'll get back to that at some point.

So 77 - > eventually one weekend I was like, know what I'm gonna do, 78 - > I'm gonna open up the Z Shell documentation from the Z Shell 79 - > website, I think was on SourceForge or something, and 80 - > I'm gonna go through line by line and try to like add some 81 - > documentation, some comments, inline comments to ex so I would 82 - > understand what all these different things did, And then 83 - > as I started doing that, I was like, oh, I had to kind of 84 - > organize this file a little bit more.

85 - > And then I was like, oh, I could use Git for this. I should 86 - > probably track my changes so that if I break something, I can 87 - > revert back to what I how I had had it originally. And so I 88 - > created a repository and then I started breaking up that really 89 - > large messy file into a bunch of smaller messy files and I 90 - > slapped the name on it and then I put a little read me and said, 91 - > here's how to install it, which is basically a git clone of the 92 - > that repository on my on my Robby Rails, I called it Robby 93 - > Robby Russell GitHub profile, it's called Omaze Shell, I had 94 - > another project like a month or two before that called Omaze 95 - > Science, and so I just kind of played it, riffed off the name, 96 - > didn't think anything of it.

I shared with my coworkers. Those 97 - > few developers that were on my team that were reluctant to 98 - > install to use it, installed it. 99 - > Great. So problem was solved.

Now I could go to the computer 100 - > and start using all the shortcuts to start showing them 101 - > things. Then they wanted to start adding their own shortcuts 102 - > and aliases, and so they're like, oh, cool. How do I do 103 - > that? 104 - > So I go, yeah, you can just, you know, we can just collaborate on 105 - > this project.

And then one of my coworkers said, how do I change 106 - > the colors? And I was like, why would you wanna do that? He's 107 - > like, I don't like your color choices, and I'm like, I have 108 - > good color choices. So I was like, well you can change this 109 - > one file, you can here's how you can change the colors, so we 110 - > started playing her with the file and started making changes, 111 - > which is great.

112 - > And other coworkers on my team started contributing, you know, 113 - > I'd mentioned other functions and shortcuts to it. Well, that 114 - > one one developer that changed the colors had a conflict in his 115 - > git repository. Because he was wanting to get the other 116 - > coworkers shortcuts, because we had a bunch of like per project 117 - > things and a bunch of little shortcuts for us as Ruby on 118 - > Rails developers and some CLI tools. And I was like, well, 119 - > this is a problem.

He he kept having to stash his changes and 120 - > then reapply them to keep his colors, and like the tweaks he 121 - > wanted to make. 122 - > And I was like, this is interesting. Well, how about we 123 - > can I'm gonna do the most obvious thing. How about we just 124 - > separate your file, terminal, your color changes from my color 125 - > things, and I'm gonna call these themes, I'm gonna create a 126 - > directory called themes and I'm gonna move mine into there and 127 - > you can have yours in there, and then we're not gonna step on 128 - > each other's toes if we make changes to our own theme, and 129 - > you can just set this as a config variable and it would 130 - > just load the appropriate file.

So that's how themes came about. 131 - > It wasn't like I planned, let's build a thing with themes, it 132 - > was literally like, how do we make this work so that we can 133 - > keep collaborating on this? 134 - > So that's kind of how the project kinda got started, and 135 - > then I was pretty reputable and well known developer in the Ruby 136 - > on Rails community at the time, And so I shared this on my blog, 137 - > robbyonrails.com, and said, hey.

Here's how I'm using all my z or 138 - > z shell. Here's my config thing if you wanna use it. Take 139 - > advantage of it. I have a bunch of Rails stuff in there, Ruby 140 - > things, stuff like that, stuff for managing servers, stuff like 141 - > that as well.

142 - > And a bunch of people started contributing to it. I'm like, 143 - > great. More themes. Everybody wanted to kind of play around 144 - > with it.

And about a month, month and a half, I think into 145 - > it, someone reached out and said, hey, this is really cool, 146 - > but it kind of assumes you're a Ruby on Rails developer, but I 147 - > work with Python, I use I've code with the Dingle projects, 148 - > how can I disable all this Ruby on Rails stuff? 149 - > And how can I contribute some Python things? And I was like, 150 - > oh, that's interesting. How about we'll I'll move my thing, 151 - > my code for Ruby on Rails stuff into, we'll call it a plugin.

152 - > You can have your Python plugin, I'll have the Ruby on Rails 153 - > plugins, just load it and we can like configure which ones you 154 - > load up, and not everybody has to have everything loaded up, 155 - > that's how plugins were started. So it was literally just me, how 156 - > do I work on my coworkers computers, teach them a few 157 - > things, they started getting excited about it, Everybody just 158 - > wanted to start contributing to it, and then themes and plugins, 159 - > I think that most people know about it, were things that just 160 - > organically came about because we needed to find a way to 161 - > collaborate, but not step on each other's toes.

162 - > And that just kind of blossomed into now it became like a huge 163 - > playground. So like, let's throw more plugins into this. And I 164 - > was like, great, I don't need that one, but come on in, let's 165 - > have fun. Let's see what happens with this.

And so it kind of 166 - > blew up over a couple years from that point. 167 - > So that's how the project got started. Anybody could have done 168 - > this. I guess I just happened to be the one to do it at the point 169 - > at that point in time.

170 - > Jack: It's such a cool story. And I I don't don't think that's 171 - > true though. We'll push back on that. I felt like you Okay.

Do 172 - > do you think it's like just I don't know. 173 - > Like it seems like that's actually the best way to start a 174 - > successful project is just like solve little problems that 175 - > you're seeing for your coworkers and stuff like that. 176 - > Robby: I always think back and, you know, I I can share that 177 - > like, I had many projects that I worked on that I had released 178 - > and open sourced, some of them if you knew how to find them, 179 - > you might find them on source forge or something, except from 180 - > like the early 2,000, horrible code, nobody ever used these 181 - > things.

What was it about this particular thing, I think we can 182 - > we can dig into that I think, But I think there was a certain 183 - > element of, I'd never set out to think that this is gonna be a 184 - > popular widely used tool. Like I literally wanted like eight of 185 - > my coworkers to have this on their computer, so that I could 186 - > be lazy and not have to remember all these complex Git commands. 187 - > And I honestly thought that that if they would take advantage of 188 - > it, they would save themselves some keystrokes and they would 189 - > find some optimizations in their workflow, and stop trying to 190 - > look up the documentation for some gits sub commands as often 191 - > as they were.

So I was like, how do we how can I just help my 192 - > team out a little bit? 193 - > So I I really wanted to kind of focus around that, and I just 194 - > shared it publicly and it kind of spread, you know. It wasn't 195 - > it wasn't like a fast burning thing where like, it just caught 196 - > white, you know, took a while for it to kind of build, and 197 - > there's like a couple of inflection points I always think 198 - > back, I'm like, oh, that's really weird. How did it get 199 - > from here to there?

You know, I have like a lot of great data 200 - > when it comes to open source projects, like how many people 201 - > have actually installed the project because the installation 202 - > is literally a git clone, and GitHub doesn't really have 203 - > really good stats on things like that. But it's it's just one of 204 - > those kind of weird things. 205 - > So I I don't know if it what the secret ingredient was. I think 206 - > there's a lot of, like maybe I was part of that, but I do think 207 - > there was a lot of good timing in the market, if you wanna call 208 - > it that.

And maybe I think maybe a little bit of, like, the 209 - > playfulness, and it was and like the how I thought about who the 210 - > target audience was for this type of tool. So we can we can 211 - > drill into those things. 212 - > Jack: Yeah. I mean, I it sounds I don't know.

It it just seems 213 - > to me like the the coolest way to start projects is just yeah. 214 - > Just build it because you think it's cool and put it out there. 215 - > And like, if you just do that, I feel like some stuff is gonna 216 - > probably be relatively popular, and in your case, very very 217 - > popular. 218 - > Robby: It can be.

I mean, there's a lot of people that 219 - > have really great ideas and then it doesn't spread. And I don't 220 - > know, you know, what what the secret sauce is for that. A 221 - > couple years ago, it was the fifteenth it was fourteen, 222 - > fiftieth anniversary, I gave a couple talks at a couple 223 - > conferences about this and, like, looking back, I'm like, 224 - > what was it that I think contributed to the success? And 225 - > I do think timing, I think was the big part of it.

226 - > Git was still relatively new. And it kind of optimized around 227 - > Git. You use Git to install it, and it was kind of targeting for 228 - > if you're getting your head wrapped around Git, I mean, if 229 - > something is simple, like, it just seems like such a simple 230 - > thing now, but the idea that you'd have a prompt that would 231 - > show you your local branch and kinda put that forward, like, it 232 - > just was making people feel a little bit more comfortable in 233 - > the CLI because what I observed was a lot of developers, 234 - > software developers that were kind of nervous opening up their 235 - > terminal and typing things in and being afraid to break 236 - > things.

And and even today, you know, know, I I I think a lot 237 - > people thought, like, well, maybe this whole terminal thing, 238 - > we will have a lot of good GUIs for doing git commands and, you 239 - > know, merging code and stuff like that. But I think a lot of 240 - > software developers still kinda gravitate towards CLI tools now. 241 - > But at the time, everybody's like, well, this is only gonna 242 - > be a short lived thing, and there's gonna be some especially 243 - > if you're coming from like the Microsoft world or something.

244 - > The you kind of assumed that there'd be this like, oh, we're 245 - > gonna do diffs in like a an app, and we're gonna be merging 246 - > things like that there, and and or even get PRs and stuff was 247 - > kind of like a new concept at the time. So I do think there 248 - > was an aspect of it being a new paradigm Git. GitHub, I knew the 249 - > people like like some of the people that started GitHub, they 250 - > thought the project was cool. They one of their they had a 251 - > podcast, they mentioned OMIZ Shell is kind of like a new cool 252 - > tool they should check out, so they mentioned it in like in 253 - > their podcast and on their blog, and then it quickly became one 254 - > of the most starred projects and remained the one of the most 255 - > starred projects for many many years, I think up until the last 256 - > several years, and there's a lot of different types of projects 257 - > that have been but it always would rank really well as a as 258 - > and and and it was always trending even though it was kind 259 - > of like not a was it funny.

260 - > It was always trending under Bash because Bash was a category 261 - > in Git, like, for different, like, programming languages. 262 - > It'd be like Ruby, Python, and Bash. And I'm like, it's not 263 - > even a Bash project. It's a Z Shell project.

If it's a it's a 264 - > shell like, I didn't know who to email there. 265 - > It'd be like, you should make that like a terminal CLI tool 266 - > type thing. But it was always trending, and I think people 267 - > would see that and be like, oh, maybe I'll check that out as 268 - > well. So just kind of like kept feeding itself in in a weird 269 - > way, and people it kept getting on people's attention.

And then 270 - > I've I I did have some fun with like how I thought about, I'm 271 - > gonna air quote, marketing the project in the sense of it's 272 - > this free tool, you know, and there's no monetization strategy 273 - > behind this. There's no company trying to make a bunch of money 274 - > off this project. 275 - > I've been selling t shirts and hats and stickers for a good 276 - > maybe twelve years now. We get almost an order a day for those, 277 - > and like so that's my monetization.

I I'm not raking 278 - > in the money, I'm not gonna buy a car with that or anything, But 279 - > that's a kind of like a fun, like, side effect of it. But 280 - > because it's just there, people keep, you know, telling their 281 - > friends to recommend it, and and it's and it also has that aspect 282 - > of like, one of the most common things I learned over the years 283 - > was, like, asking people this, like, oh, I love I use OMXD 284 - > Shell. The question is, like, how did you hear about it?

285 - > I always would ask, who introduced you to it? Because I 286 - > feel like a lot of the time the stories I hear is actually 287 - > someone's looking over at someone's computer and like, how 288 - > did you make it look like that? Like, oh, check out OMAZ Shell. 289 - > And that was the story I kept hearing over and over and over.

290 - > Like, coding schools would tell their their new students like 291 - > the boot camps, like, oh, just install OMAZ Shell, like day 292 - > one. 293 - > One of my one of my really, really good friends personally, 294 - > a couple like two years ago had got a new job somewhere, and he 295 - > was reading through the documentation for getting 296 - > onboarding, and it's like, your name popped up in our our 297 - > Confluence documentation. It's like, install Robby Russell's 298 - > only Z Shell day one.

And I'm like, these things is like it 299 - > just I don't know how how that's all happened. It's just like 300 - > this wild thing. So it wasn't a strategy that I felt like, hey, 301 - > everybody, you should all make this the default. 302 - > And and then also Mac or Apple at one point ended up switching 303 - > from Bash to Z Shell as a default for a open source 304 - > license reason, because Bash and Z Shell have a different 305 - > license, and so that ticked up the popularity a little bit more 306 - > as well, and which has also added some interesting 307 - > confusion, because some people don't know what the difference 308 - > between Z Shell and Omey Z Shell is.

I don't know if I can solve 309 - > that problem, but it it has been like a thing, people are like, 310 - > well, it's just is it the same thing? And like, my thing's 311 - > built on top of it. It's kind of like explaining what Ruby on 312 - > Rails is versus Ruby if you're not a software developer, people 313 - > are like, what exactly does that mean? It's kinda like a similar 314 - > thing where they're like, well, it's kinda the same thing, 315 - > right?

316 - > And they're like, no, not quite. But, and I know that I've heard 317 - > rumors that there's people in the Z Shell support community 318 - > that don't like Oh my Z Shell because people confuse it, and 319 - > they'll come to them with support questions that are 320 - > actually probably for a normal Z Shell configuration problem or 321 - > something. So I I apologize for that. But but at the same time, 322 - > maybe I this project helped introduce a lot more people to Z 323 - > Shell in the first place.

324 - > Jack: It's super cool. My friend Tony, I think introduced me to 325 - > it a long time ago and it's all like the memories are all hazy 326 - > at this point but I remember I the felt like the branching was 327 - > like a really big thing where I was like, well, how do you how 328 - > can you see that? What which branch you're on? Are you not 329 - > doing like like yeah.

It's it was I I think that was like 330 - > really cool and it just looks really nice. 331 - > Robby: Thank you, Tony. 332 - > Jack: I need to catch up with that guy. Great guy.

Yeah. It's 333 - > a it is really cool hearing about this. And then were there 334 - > like a lot of I guess when you build a popular project, it's 335 - > like you kind of have to make a lot of decisions around like how 336 - > it develops because people often want like different things. 337 - > I mean, like you can see it now on like Twitter with like 338 - > OpenClaw, like the guys like getting a gazillion requests for 339 - > things and stuff.

And like, I can imagine like it was very 340 - > much like you were getting tons of requests and like Yeah. How 341 - > are you thinking about like developer experience and like 342 - > deciding what to build and what not to build and stuff like 343 - > that? 344 - > Robby: It's a topic that I have given a lot of thought about 345 - > over the years. And I mean, initially, my target audience 346 - > was my team.

Right? And so I always and my team has evolved 347 - > over the years, and we, you know, but we would always have 348 - > you know, back in back in that era, so this is in 2009, what we 349 - > we used to call we used to have a type of developer, called what 350 - > we used to call a front end developer was kind of someone 351 - > that knew a lot of HTML and CSS, maybe a little bit of 352 - > JavaScript, but we could work on our projects and bring our 353 - > design, like our our clients like, designs and bring them 354 - > into our Ruby on Rails apps and make sure that they work well 355 - > across browsers and different devices and stuff like that.

So 356 - > that was our front end developers, what we used to call 357 - > those people back in the day. 358 - > Front end developer, and it means different things now. But 359 - > they were my litmus test. If that developer who was very 360 - > uncomfortable in the CLI and maybe a little uncomfortable and 361 - > still new to git, if they felt comfortable enough to use it, 362 - > and then I would feel like, great.

Then this is kind of 363 - > that's the litmus test. Like, I wanna make sure it worked well 364 - > for them. 365 - > And I thought about that when we would when you know, when the 366 - > project started to grow more and people started adding new 367 - > functionality. Like, when you bring in like themes and 368 - > plugins, a lot of that was like, is there some good useful 369 - > documentation on how to use it?

Great. If they don't need to 370 - > know how to use that plugin, it's not a big deal because it's 371 - > those are things you opt into. You would enable plugins. 372 - > They just weren't all gonna be loaded.

They would come they 373 - > would get installed by you know, when you installed Domain Z 374 - > shop, but like you don't need to know that there's a Django 375 - > plugin if you're not a Django developer. Someone would tell 376 - > you or you would know to enable these things. But when it came 377 - > to like all the git stuff, and how you're gonna keep it 378 - > updated, I wanted to really optimize around just that type 379 - > of developer. And so early on, I thought about like, we were 380 - > already having issues as people were making changes, like how do 381 - > you keep the how do you stay up to date with this project when 382 - > it's a git repository?

383 - > And so like, they wouldn't remember to c d into that, you 384 - > know, the .omyz shell directory that you would have in your home 385 - > directory and then do like a git pull and fetch nude. And so 386 - > they'd be like, one coworker be like, hey, added this new these 387 - > new little features and the other ones like why it doesn't 388 - > like not it doesn't it doesn't it's not working for me. I don't 389 - > see it.

And I'm like, oh, you need to like go into the 390 - > directory and do a poll, and then it would fetch that. They 391 - > wouldn't be thinking about that because it's not because it's 392 - > just like their CLI interface. 393 - > It's not like working on a client project or a software 394 - > project with your coworkers, and everybody's like, you know that 395 - > that's what you need to do on a regular basis. But since it was 396 - > kinda like out of mind, how do you think about that?

And so I 397 - > was like, wow, this is an interesting thing. So I was 398 - > like, how about is there a way to automate the in like the 399 - > updates? And like all I wanted to do and so I think the initial 400 - > version would be like, maybe once a week, I had to figure out 401 - > how to do like a epoch time thing in Z Shell to like track 402 - > when the last time you ran an update, if it had been like over 403 - > seven days, it would then ask you, do you wanna do an update?

404 - > Yes or no? And so I was learning how to do some Z shell concoding 405 - > through this process. And so like, I'm like, oh, I'll make 406 - > this an update auto update thing and you can turn it off if you 407 - > want. But that would make it easier because then that front 408 - > end developer would just have to be like, yes.

You know, I type a 409 - > y and then it would just fetch the new updates and get back to 410 - > what they were doing. 411 - > And I'm like, great. Then they can stay up to date. You know, 412 - > did you update?

And there was like a little, you know, like a 413 - > little OMZ update command you could run and then you could 414 - > just get so that way if something was more recent in the 415 - > week, you weren't too far behind. And then that be I think 416 - > that was like a secret ingredient to the project 417 - > globally though, because I think because then it was like, oh, 418 - > now everybody has is by default is going to be relatively up to 419 - > date with new things showing up.

420 - > And because it was git, it would show you when you ran the when 421 - > the update would run, here is new plugins. You would get 422 - > exposed to new themes, new plugins that were that were 423 - > added. We didn't have to have like a huge complex like, go 424 - > check the change log and see what's going like. So the 425 - > approach was like, well, let's just lean on git as much as 426 - > possible.

And then I think at some point, was like, oh, I can 427 - > add like the ASCII art in there, and it'll be kind of fun, and 428 - > it'll be like a promo, oh my Z Shells, like there's there's 429 - > this community of things getting added, You can if you're 430 - > interested in it, you can go look into it, but you're 431 - > probably just gonna go about getting on with your the project 432 - > that you're working on. 433 - > But it always felt like that way, it always felt like Omaze 434 - > Shell was kind of like there, but out of the way.

But in terms 435 - > of like thinking about that type of developer in particular, 436 - > someone that's not super comfortable with the CLI, within 437 - > maybe a couple years, we had a couple of people that were 438 - > really really helpful in the project. And they're still I 439 - > think ranked probably in the top fifteen, twenty contributors to 440 - > the like, code wise to the project. And we started working 441 - > around a lot of things to make it faster. And so we're we were 442 - > trying to optimize and speed up thing.

443 - > But then the plug in ecosystem was growing pretty wide bigly 444 - > you know, bigly Growing pretty big, and we had hundreds of plug 445 - > ins or something. And I was like, well, this may not scale 446 - > well. Like, we do we need to bring like, let's talk about a 447 - > plug in system. So we we started exploring this, and we were 448 - > trying to approach like all these different ideas about how 449 - > we could do that, and thinking about themes as well.

Like, 450 - > maybe we need to put a cap on how many themes are, because if 451 - > someone's upload you know, adding a theme, and it doesn't 452 - > look that much different than like 20 other themes, are we 453 - > actually helping the community by adding a bunch of new themes 454 - > every time? 455 - > I'm like, well, people can have their own custom themes. If 456 - > something really innovative shows up, maybe we'd consider 457 - > that. So we had to like kinda put a lockdown on things like 458 - > that, so it didn't get too big and bloated.

But when it came to 459 - > like the plugin system, we're like, how will we organize that? 460 - > And there was a couple patterns and we started going down this 461 - > path of using, which some developers might be listening to 462 - > this be like, oh, please don't do that, was to use git sub 463 - > modules, like let's lean in on git more and we'll use git sub 464 - > modules. 465 - > To do this, and as we were trying to like think about how 466 - > to make this easy for developers, I was just like, you 467 - > know what, GitHub bones are actually kind of really messy 468 - > for even me to use on any of the projects I've ever had to use 469 - > them.

I do not wish that on anyone else, like if we have to 470 - > do if we go down this path, I think it's gonna make the 471 - > barrier of entry very very complicated. So I had to be 472 - > like, we're not doing that. I'd rather keep plugins as as yeah, 473 - > let's add more plugins, we'll keep doing this, but I'd rather 474 - > have that problem and how we can manage that than to make it too 475 - > complicated to use for these types of people that are not 476 - > comfortable with the CLI.

And if they have to run some extra git 477 - > commands that they don't understand what they're doing, 478 - > we're gonna lose them. 479 - > And it's not necessarily that where I was worried about losing 480 - > people, but I was just like, I'm like, I think if you're 481 - > sophisticated enough to to use to to use those types of 482 - > functionality, great. Maybe you can graduate beyond OMXZ Z 483 - > shell. You can go do these things.

You don't need OMIS Z 484 - > shell anymore. So it's like, how about we just think of this as 485 - > like the tool for people getting comfortable with the CLI. 486 - > And then when you're ready to graduate or whatever, you can do 487 - > that and you can figure out how to do that stuff. So we won't be 488 - > for the advanced developers, we'll be for like the newbies, 489 - > the beginners that are kind of excited and like, yay, I'm a 490 - > programmer now and I keep my terminal looks like, I feel like 491 - > a hacker or whatever, I'm like, you know, I've got these cool 492 - > colors and I'm learning how to do all these little cool 493 - > shortcuts and using these plugins and plant switching my 494 - > themes around.

I'm like, but like that that seemed like a 495 - > more exciting and fun group of people to cater towards, and 496 - > then knew that like the curmudgeony people that like, I 497 - > want it to be super fast and I wanna do it my way, great, go do 498 - > your thing. I don't we don't need to be that for you. And 499 - > what I found was that, yeah, people will graduate, but 500 - > there's all I meet so many people that are like really 501 - > really smart savvy developers. 502 - > So yeah, I just install all my Z shell when I get a new computer, 503 - > make a few little quick tweaks and I just it it works.

I don't 504 - > I actually don't care for a more complex, and I'm not looking to 505 - > shave off a couple milliseconds of my my prompt startup time. 506 - > And I'm like, great. I'm glad you stuck around as well. So 507 - > that that's that's how I thought about the developer experience.

508 - > Like, that's the target audience. And it just happens to 509 - > be that there's a lot of them because people when people are 510 - > new to something, like, let's optimize around that space. 511 - > Jack: Yeah. Was it like instantly obvious to people?

512 - > Like, did you have to like communicate? Because I guess 513 - > people 514 - > Robby: The forked the project got forked. 515 - > Jack: Okay. 516 - > Robby: When when when I when I when I made that decision, and 517 - > we can maybe we can go find like the the GitHub PR or issue that 518 - > that was on, the project ended up getting forked.

And there's 519 - > there was a fork that still I think it still exists to this 520 - > day that started going down that path. They didn't fully go 521 - > through that plan that they had worked around, but that was 522 - > there was a pivot point where some of those people said, 523 - > alright, I'm out. I'm gonna go do my thing. And I was like, I 524 - > understand and that's okay.

525 - > So the beauty of open source, you know, wasn't like bad blood 526 - > or anything, and then other people started kinda came in and 527 - > filled in that void and from a contribution perspective, and it 528 - > worked out. And so it wasn't like, I wasn't ungrateful for 529 - > their help, but it was like, I'm like, I'm like, we're we're just 530 - > different types of developers, like you, you're way smarter 531 - > than I am, than I than I think I can wrap my head around this.

532 - > I'm like, I don't know how to communicate your ideas simply to 533 - > somebody coming into this new, and they're like, well, that's 534 - > not our problem, and I'm like, I don't, I'm not trying to be 535 - > discouraging of them, it's just, if you know, if they're out 536 - > there listening, but it it was it's just there was still like 537 - > a, how do we just I just wanna focus on these types of people, 538 - > I'm like, let's let's see what happens, and then that's gonna 539 - > be my bet on it, and I'm like, that's where I'm kind of excited 540 - > about it, and and I I didn't think this project would exist 541 - > fifteen years later, you know, but it but it has.

542 - > And it wasn't like I thought this was a good long term 543 - > strategy. It was like, I always thought OMAZZ Show would be 544 - > replaced by some new exciting things that did it even better 545 - > and like, great. And people I I was chatting with somebody 546 - > recently. I get this a lot too.

It's like, not only like, oh, I 547 - > love OMAZ Show, but it's like, I'm sorry, but I actually 548 - > switched to fish like two months ago. 549 - > And like, they're kind of apologetic, and I'm like, what 550 - > are you talking I'm like, that's great. I'm glad you found 551 - > something for you. And like, it's not, you know, I I still 552 - > use all my Z show.

I haven't been motivated. I I'll dabble 553 - > with different tools, but I don't feel like I need to get 554 - > into like a Vim versus Emacs debate about these things 555 - > either. 556 - > So find what works for you and what resonates. 557 - > Jack: Yeah.

It's it's so cool. And, like, how actually, just 558 - > like a cure question just for my own curiosity, like, how what 559 - > what are like the biggest impacts on your life? Like is 560 - > did it change things? Because building such an popular open 561 - > source projects, obviously, it's not like a business venture, but 562 - > it's like a kind of something that's incredibly popular.

563 - > I just wondered, like, did it have a lot of impact on your 564 - > life and work and stuff like that as well? 565 - > Robby: It's interesting. I some people think that it probably 566 - > had more of a boost to my professional career than I think 567 - > I think I it's it's hard to know. I mean, we've gotten my 568 - > company because we were I run a software consultancy and been 569 - > doing that for much longer than know, this all kind of brewed in 570 - > the that that world is that companies my company still 571 - > exists.

And OMAZ is still kind of seen as like Robby's weird 572 - > little project, you know, within my company. Like like, I think 573 - > most of the people on my team use all my Z Shell, there's a 574 - > few that definitely don't, I know, and they're very vocal, 575 - > like, I don't use it. 576 - > I'm like, I'm not gonna we're not pair programming in the same 577 - > office anymore, so I don't really care about what what 578 - > they're using anymore, because I'm not gonna be on their 579 - > computer typing myself, but like I needed to like sixteen years 580 - > ago.

But professionally, I, you know, I sell you know, I 581 - > mentioned I sell hats and t shirts and stickers all the 582 - > time. That's been fun because a from a marketing perspective, 583 - > there's a much wider audience of people to not necessarily to 584 - > sell things to, but to experiment with like marketing, 585 - > promoting a project. It's it's a huge community and so I I feel 586 - > like that's been fun over the years and to to play around with 587 - > that. It definitely has helped introduce me to a lot of people 588 - > I look up to.

589 - > I think it helps open up doors for, like, my my couple of my 590 - > podcasts. I have a podcast called Maintainable. And on that 591 - > podcast, I think it's helped. People are like, oh, people that 592 - > have used OMXZ Show and but they're like, they've written 593 - > this cool book and I'm like, I just wanna get to talk with them 594 - > for an hour about software maintenance.

And they're like, 595 - > oh, I would love to chat with you because you're the OMXZ Show 596 - > guy. 597 - > And I'm like, great. So that's I think it's been helpful from 598 - > that perspective for sure. But from like a financial 599 - > perspective, it's I can't say like we'll have clients that 600 - > don't know anything about Omezuchel and this is not even 601 - > remotely on their radar of like, it it occasionally does pop up.

602 - > Well, I've had some clients where they'll be like, oh, I was 603 - > talking to one of our other internal software developers, 604 - > and they they're like, you're famous. You know? 605 - > They're like but they don't know why I'm like, the my client 606 - > doesn't understand why I'm, you know, recording for our audio 607 - > listeners why I'm famous for doing this thing. But I think it 608 - > that that part's been interesting.

And the other thing 609 - > it's been the most rewarding, I think, is just being able to 610 - > contribute to a to get to participate and, you know, like, 611 - > while I created OMNIZZ Show, I've kind of relabeled that as I 612 - > I feel like the better word is curated OMNIZ Show in a way. 613 - > Because, again, it was like I compiled all these ideas from my 614 - > peers, my friends. The project, I think, were just shy of 2,500 615 - > contributors to the project that have committed code to the 616 - > actual repository at this point, which is a lot of people that 617 - > we've reviewed and approved code for.

618 - > And we are primarily managed by three maintainers right now. And 619 - > there's two, Mark and Carlo, who both live coincidentally, like, 620 - > completely coincidentally. They live, like, within two miles of 621 - > each other in in Spain. And and so and and we're all like ten 622 - > years apart in age.

So I'm like in my, you know, I'm like 46, 623 - > the other one's like 35, and the other one's like 25. 624 - > And like, we have this age range of people, and like, the 625 - > project's almost as old as Carlo is. You know? And he's one of 626 - > our most active, you know, maintainers of the project these 627 - > days.

And they're the ones doing a lot of the PR reviews and and 628 - > doing some of the longer, the bigger the bigger, like, 629 - > refactoring projects and stuff like that. But so it's it's 630 - > allowed me to meet so back to your question, I've gotten to 631 - > meet a lot of interesting people because of the project. 632 - > I'll show up at conferences specifically. I'm, you know, I'm 633 - > still in the Ruby on Rails community.

I host a Ruby on 634 - > Rails podcast, and people know me from the Rails community. But 635 - > more people at those conferences know me because of OMIZZ Shell 636 - > than they know that I am actually a Ruby on Rails 637 - > developer, and that's actually how I make a living in as a 638 - > software developer. I wish they knew me more for that, so we can 639 - > get more clients for that. 640 - > But it's just like a funny, like, so I'll try to so I'll 641 - > lean in on that.

I'm like, hey, everybody, if you use all my 642 - > Zisha at the conference, I always got stickers on me, come 643 - > say hi, and then I can use as an excuse to get to know them and 644 - > everything as well. And so it's it's a kind of a funny thing 645 - > that I wish I'd figure out how to optimize it around my 646 - > professional stuff more. Or I've always been like, oh maybe I've 647 - > missed a missed something because I didn't figure out how 648 - > to monetize this project.

But I'm also like really okay with 649 - > that because I've always felt like that I don't owe anything 650 - > to the broader community. 651 - > It's like, I'll work on it when I feel like I'm interested and 652 - > have some energy, but it might kind of take on managing a 653 - > project of this scale. It's like, I always felt like it was 654 - > feature complete day one. It's just that other people wanted to 655 - > start changing my thing and they wanted to customize or add a 656 - > couple of extra little things.

And my thought there is like, 657 - > unless there's a security thing or a bug that's like a lot of 658 - > people they're being impacted by it, and that rarely happens, but 659 - > when it does, like those are those definitely, you know, 660 - > bubble up to the top. But if you're a you if you're a first 661 - > time contributor and you've got like a new plugin idea, you've 662 - > figured out how to fork the project, add this new plugin, 663 - > start working on it, add documentation for it, unless a 664 - > bunch of people have already seen that issue or the PR and 665 - > are upvoting it, nobody's waiting on it.

666 - > It's gonna be a new thing that we'll get around to, but you're 667 - > able to use it because you've already you're smart enough to 668 - > know how to already go through that step that you have that 669 - > plugin enabled in your local fork of it, and so you're not 670 - > stuck waiting for me to review and approve that. So we'll get 671 - > to your PR when we've got time, because we're all volunteering, 672 - > and that's kind of how I think about the project from like, so 673 - > people are like, wow, you've got hundreds of open issues or PRs, 674 - > I'm like, yeah, but is anybody waiting on them, really?

I mean, 675 - > just the person that wants to get the contribution merge, And 676 - > like, so it's like, there's a few people, I'm not trying to be 677 - > dismissive of that, but it's just like, if we did it quicker, 678 - > then I'm sure it would just create more PRs for ourselves. 679 - > So try to find a good healthy balance there, and keep it like 680 - > this is a volunteer hobby when it when I'm when I feel 681 - > motivated to work on it, and not feel like it's like I have to 682 - > spend x number of hours each week on this particular project.

683 - > The other thing I would say though, has this benefited? I 684 - > actually do know that, like Mark and Carlo, because they've been 685 - > working on this project for a number of years as well, I do 686 - > know that that's helped open up doors for them professionally. 687 - > More so, I think it's been more applicable for what they've been 688 - > doing. So then when they can show, like, they're kind of a 689 - > you know, at one point, you know, the one of them had just 690 - > finished university, other one had finished it several years 691 - > ago.

Like, to get some of those entry level, get your foot in 692 - > the door things, like, to be able to say that they've worked 693 - > and been a maintainer of Omeizusha, I think has helped 694 - > open up doors for them. 695 - > And that so that's that's that's pretty rewarding to to know that 696 - > that that's helped helped them in particular. 697 - > Jack: Yeah. One transition question here.

Have LLMs made it 698 - > I'm guessing this make it harder to, like with, like, poor 699 - > requests and stuff. Chiggo, have you had, like, a massive influx 700 - > of them? 701 - > Robby: We've had we've had a handful of them periodically 702 - > come in, and it's been it was an issue several months ago that we 703 - > needed to have a couple conversations about it. And as 704 - > the maintainer team and like, say like, alright.

Well, we 705 - > should probably update our contribution gains to talk about 706 - > AI more specifically. And so we we just pushed that out maybe, I 707 - > think about a month ago. I I made updates to our contributing 708 - > documentation for that, and just to be like, alright. 709 - > What the what the team really essentially wanted was like, 710 - > hey, if we if we get these big PRs, we feel kinda bad to say no 711 - > if we think it's AI.

We wanna they wanted to feel comfortable 712 - > to be like, we're just not gonna consider this thing because it's 713 - > maybe it's too sprawling. We've seen a lot we've seen some PRs 714 - > where it's just like, they touch like 25, 30 files, and we're 715 - > like, what what are what are we doing? What is this for? You 716 - > know, like, makes us a little nervous.

717 - > But I mean, I'm like, but when we thought about it, it was 718 - > like, if that hadn't even been in just a normal human, like, 719 - > just a human on their own, we would have a very similar 720 - > reaction. We've yes. We've seen an increase in AI supported 721 - > assisted PRs and contributions, and we decided we wouldn't rule 722 - > them out immediately and just say, like, no contributions. We 723 - > just said, we ask people to disclose it because we don't 724 - > know if people can say, did they have it generate the whole thing 725 - > or did they use, like and if you're using Versus Code and 726 - > then there was, like, an autocomplete, does that count as 727 - > using AI assisted?

And, like, where's that line? 728 - > I feel like the tools are gonna keep evolving, and you know, I 729 - > didn't wanna have to revisit the policy every six months, and I 730 - > don't feel like I'm in a position to be like, I'm gonna 731 - > have an ethical stance of whether or not AI should should 732 - > or shouldn't be used in this open source project. I'm like, 733 - > but I can tell you that because I use Copilot to make a copy or 734 - > a find and replace for a few links in the the project, that 735 - > resulted in a few people saying they weren't gonna use the 736 - > project anymore because it's full of AI slop now.

And I'm 737 - > like it's like a we're we're we're navigating, I think, an 738 - > interesting, coding cultural war there. 739 - > Jack: Super interesting. 740 - > Robby: Right. We're we'll see where this goes in the next few 741 - > years, I guess.

But. 742 - > Jack: And you're seeing a lot of I think you mentioned that you 743 - > had, like, some fairly strong viewpoints on what you're seeing 744 - > with AI and your consulting work, where it's useful, where 745 - > it's not. 746 - > Robby: Well, I think like every two months, I become becoming 747 - > more begrudgingly, like accepting that my type of 748 - > business model may be doomed. And that's okay.

Or at least the 749 - > the way it historically has worked and trying to, like, be 750 - > open minded enough and then be like, well, maybe things do need 751 - > to shift and, like, things will be different. And how do we fit? 752 - > And what does my business model what does it look like in two 753 - > years from now? 754 - > And it's feels very murky right now.

I But think a lot of 755 - > software developers really like this. I don't feel like it's a 756 - > unique thing. I'm just like, well, my business model is like, 757 - > we come in and help people with existing legacy applications, 758 - > and we're really good at taking over those projects and being 759 - > good stewards to give them like a second wind, a second act on 760 - > their projects. That's what we specialize in.

761 - > AI might make that a lot easier for a lot more people to do 762 - > that, and which is great. I want because my big thing is, like, 763 - > I'm very much a strong believer in not rewriting, but I'm 764 - > starting to feel like maybe that argument because I always feel 765 - > like it was a huge waste of the human effort that went into 766 - > those projects to just be like, we're just gonna rewrite this 767 - > thing because we also know that rewrites are often way 768 - > underestimated.

They're way more complicated, and you end 769 - > managing multiple applications in parallel until you get to 770 - > fully do a cutover, if that ever even happens, and then people 771 - > quit and the project doesn't ever finish anyways. Yada yada 772 - > yada. Like, all those things are the realities of software 773 - > rewrites. 774 - > But but maybe software rewrites might be easier now with the 775 - > tooling, but I also still don't know if that's actually the best 776 - > solution there.

So so I I do think we can find our our niche 777 - > in the market and and and stay in business there. But it 778 - > definitely things are different now. And I don't and we're and 779 - > I'm finding reasons to rewrite things myself more often than I 780 - > used to. 781 - > Jack: I mean, are you beginning to see the things that you might 782 - > focus on that like would have a would be, like, more resistant 783 - > to AI, I guess?

784 - > Robby: I the thing that I am gravitating towards is, like, a 785 - > lot of the types of software projects that we've really 786 - > enjoyed working on over the years have typically been some 787 - > really small focused internal operations tools for 788 - > organizations, where they're not the business itself. Usually, if 789 - > someone's gonna build a SaaS product or a software project, 790 - > then they're gonna try to sell it and find a market for that, 791 - > probably a lot of the people that you have on the podcast.

792 - > The that's the product. Right? And so you're gonna have your 793 - > own engineering team. 794 - > But like we end up taking over a lot of projects where you don't 795 - > need two or three full time people working on it to keep it 796 - > up and running and keep it stable, you just need a reliable 797 - > partner to take care of it and tend to it and address bugs and 798 - > performance issues or integration challenges like, 799 - > hey, we have this thing that or ERP that's like integrating two 800 - > different systems and it's not working all of a sudden in one 801 - > weekend, and like then we can dig into those issues.

So I 802 - > think the more we think about just keep focusing on like 803 - > what's those like smaller tool, like software projects. And I 804 - > think there's been a trend for several years for organizations 805 - > to stop building their own custom solutions and start using 806 - > SASS because they didn't have to man maintain the thing yourself 807 - > anymore, because it's expensive to maintain software. So I 808 - > understand that. But a lot of companies started to optimize 809 - > around a SaaS product that would give you like get you kind of in 810 - > the ballpark of like say 80% of what you were looking for and 811 - > you needed to kind of like work around, that could be like 812 - > literally humans having to work around or export things from 813 - > that system into spreadsheets and work out some stuff, then 814 - > they can do another part of the process in another system, or 815 - > try to figure out integrate with those SaaS products.

816 - > And I think some companies lost some of their secret sauce as 817 - > part of how they do things to optimize around something more 818 - > generic like a SaaS product. I do think there's now an 819 - > opportunity for for for companies in particular to be 820 - > like, why don't we build tools where there's one user, it's us, 821 - > you know, like, let's build it around what makes us unique as 822 - > an organization. And we can learn patterns from all these 823 - > different SaaSes, but we don't need to rely on the SaaSes 824 - > because modeling like, oh, we want Kanban boards for managing 825 - > our internal CRM tool.

We don't need to pay HubSpot anymore for 826 - > that. We can just we can maybe we could code that with Cloud 827 - > Code over a couple weeks and and but just focus around the what 828 - > makes us special and unique. 829 - > I do think that's something that I think software consultancies, 830 - > freelancers could be thinking about, like, what makes the 831 - > company unique and focus it around that. And eventually this 832 - > might add back to like, well, we don't need to have this main, 833 - > you know, because if because that was how software used to be 834 - > built.

It was always optimized around that initial company. 835 - > They didn't they weren't interested in like turning into 836 - > a product that other people could use. Use. 837 - > It was it was a lot of vendors like mine that would build that 838 - > custom thing for multiple clients and see similarities, 839 - > and like, what if we just started building a similar tool 840 - > for a lot of companies like that?

And so, which was I think 841 - > a good business model, and like now I think it's kind of 842 - > swinging back the pendulum, maybe going back to that kind of 843 - > like, let's focus around the custom unique aspect of how this 844 - > company operates, what makes them special, unique, focus it 845 - > around them, and stop trying to just be like, well, how would a 846 - > normal generic CRM work for a company like us? Like, no, let's 847 - > just build the CRM that actually works the way we want it to, and 848 - > we can tweak it, And not feel like we're trying to figure out 849 - > to tweak this kind of generic CRM that is competing against a 850 - > bunch of other generic CRMs in in the market.

Maybe. I don't 851 - > know. 852 - > It's an idea that I've been kind of thinking a lot about lately. 853 - > Jack: Yeah.

I think it's I think it's true. Like, we're doing 854 - > that. We're using building a lot more of our own stuff and yeah. 855 - > Feels like so maybe that's, yeah, gonna be like, everyone 856 - > needs these and maintain these tools they've built and stuff.

857 - > Maybe. 858 - > Robby: The only other thing I would say there is that because 859 - > of tooling like AI, LMS, assisted coding is just like a 860 - > pattern that a lot of companies have had and we've seen over the 861 - > years is, software engineering teams will optimize their 862 - > software architecture around their essentially their org 863 - > chart, and how they perceive their org charts gonna continue 864 - > to grow. And I if we've seen over the last few years, a lot 865 - > of companies have contracted in size.

And so the engineering 866 - > teams are smaller and having to deal with the consequences of 867 - > the decisions that teams were making when they thought the 868 - > teams were going to be bigger one day. And I think software 869 - > engineers this is a good time, I think, for everybody in an 870 - > organization to think, great. We're building these tools and 871 - > we're scaling in a certain way, but would it be actually ideal 872 - > for the organization to one day have a smaller engineering 873 - > footprint?

874 - > What sort of decisions should we make for that future smaller 875 - > team that hopefully I'm part of, but there's a good chance that I 876 - > might not be. I might move on for my own professional reasons, 877 - > or the company might need to scale back the size of their 878 - > team. And maybe that's the most optimal situation for that 879 - > organization. So I think if we're thinking about scaling 880 - > things, we should also think about, like, how do we build and 881 - > optimize our software so that it can be maintained by a smaller 882 - > team in the future?

And that might be the best case scenario 883 - > for that particular organization to be stable and well 884 - > maintained. 885 - > So 886 - > Jack: Yeah. Super cool. Robby, conscious of your time.

Thank 887 - > you so much. Where can people learn more about you and what 888 - > you're working on? 889 - > Robby: Yeah. You can head up to robbyonrails.

com, and there's, 890 - > like, a there's a link. This is, like, elsewhere, and there's a 891 - > bunch of links there. But otherwise, I'm pretty easy to 892 - > find just Robby Russell on the Internet. 893 - > Jack: Amazing.

Well, thank you so much for joining, Robby, and 894 - > thanks everyone for listening. 895 - > Robby: Thanks for having me, Jack. It's been a pleasure.