macwright-com-o4dndf.txt (16371B)
1 Tom MacWright 2 3 [email protected] 4 5 Tom MacWright 6 7 • [1]Writing⇠ 8 • [2]Reading 9 • [3]Photos 10 • [4]Projects 11 • [5]Drawings 12 • [6]Micro 13 • [7]About 14 15 A year of Rails 16 17 Railroad 18 19 I spent most of 2020 working with [8]Ruby on Rails. I moved a project from [9] 20 Next.js + [10]Rust to… Rails, baby! Back to the future. My earlier post on [11] 21 Second-guessing the modern web was inspired by this experience, that for the 22 product we were building, a ‘modern’ stack was not working as well as a 23 traditional one. 24 25 We didn’t do competitive analysis against Laravel, Django, or Phoenix. They’re 26 similar, not radically better or worse. There are multiple acceptable solutions 27 to a problem, and this was more a matter of choosing the right kind of solution 28 than pursuing some kind of perfect choice and burning hours and motivation 29 doing the window-shopping. 30 31 What helped Rails win was that the team had a little more experience in Ruby 32 (with the exception of myself), and we found plenty of resources for developing 33 and deploying the stack. Rails fit perfectly into the ideology of [12]Choosing 34 boring technology. Another part of the product would be the hard, innovative 35 part, so it made no sense to grapple with bleeding-edge web frameworks. 36 37 This was a really fun experience. There’s a lot to love about Rails. Other 38 communities could learn a bit from the Ruby & Rails culture and wisdom. I won’t 39 implement everything in Rails, but it’ll be part of the toolbox. 40 41 Before this, I hadn’t touched the stuff. And I bet a lot of people are like 42 that - they came of age in the world of React and Go, and haven’t tried 43 anything even remotely similar to Rails. For their benefit, and to debrief from 44 2020, here are some notes on the experience. Plus, [13]Rails-like projects in 45 JavaScript are ramping up quickly, and it’s fun to know the origins. 46 47 The good 48 49 Debugging Rails apps is amazing 50 51 A while ago, I [14]wrote on Twitter 52 53 the real reason why javascript developers don’t use breakpoints and use 54 console.log is that breakpoints don’t work 55 56 After years of working in JavaScript, I’m used to bad debugging experiences. 57 The Chrome debugger’s [15]automatic pause on caught exceptions is amazing, 58 sometimes. But throwing a debugger statement in some React code is dodgy as 59 hell. Sometimes it works, mostly it doesn’t. You have to deal with code that 60 might not have the right [16]sourcemap to translate from bundled & minified 61 code to original source. Subtle abstractions like React hooks and advanced 62 transpiler stuff like [17]Regenerator mean that your code’s stacktrace probably 63 looks nothing like what you expect, with lots of internal garbage. Sure, you 64 can learn better techniques for diagnosing and debugging errors, but it’s not 65 just you - the debugging story in JavaScript is pretty bad. This applies even 66 to Node.js, where one of the debugging stories is to connect Chrome’s debugger 67 to a Node.js instance: a finicky solution that doesn’t consistently work. 68 69 In Rails, there is [18]byebug. You write byebug in your source code, and you 70 get an interactive REPL right there. It works in views, controllers, database 71 migrations, everywhere. It almost always works. Variables are named what you 72 expect. The whole system is paused at that moment, and you can actually 73 interact with it, using all of the Rails utilities and your installed gems. 74 75 If a page crashes unexpectedly, you get a similar REPL experience, in your 76 browser, automatically. With an automatically cleaned-up stacktrace that 77 excludes Rails’s own frames. Like the byebug interface, this REPL actually 78 works and is consistently helpful in finding root causes. Rarely will you need 79 to use puts to print something to the console because this debugging system is 80 so good. 81 82 The magic mostly works 83 84 Our Rails app didn’t have any require statements. You mention a module’s name, 85 and it’s automatically included, using [19]Zeitwerk, a tool that comes standard 86 with Rails. 87 88 This kind of system was terrifying to me before. What if you accidentally 89 import something just by mentioning it? What if two things have the same name 90 and you import the wrong one? How do you really know what’s happening? Sure, 91 you’re happy now, with all of that annoying importing and exporting taken care 92 of, but the sky might fall. 93 94 Or maybe it just… doesn’t. Maybe impure, vaguely risky techniques are just a 95 net positive over time, and making everything fully explicit isn’t really 96 necessary? Now when I’m using other systems, I wonder - what if I could just 97 mention one of my React components and it would just… be there? Sure, the 98 system would have to complain if there were two components with the same name, 99 and it would have to make assumptions about directory structure, but overall, 100 wouldn’t this be nice? 101 102 This applies to a lot of other parts of the system too. Rails is famous for 103 doing pluralization - you name a model Post and you automatically get an 104 interface called posts. But what, you ask, of words with uneven pluralization 105 rules? Rails actually [20]does the right thing, almost always. And when it 106 fails, you can override it. It actually just saves time, reliably. 107 108 Testing works 109 110 I’ve tried to test front-end applications. I’ve set up [21]nightwatch, [22]jest 111 , [23]enzyme, [24]cypress, and probably 5-10 other frameworks. Front-end 112 testing is universally terrible. Projects like Cypress are throwing untold 113 hours into making it less terrible, taking on massive amounts of complexity to 114 abstract away from fickle browser behavior and complex interactions. 115 116 But it still sucks. Frontend testing has no good attributes: it’s unreliable, 117 hard to automate, hard to debug when it fails, and often doesn’t even assert 118 for important behaviors, so it doesn’t actually identify regressions. Running 119 frontend tests in CI is resource-heavy, requiring you to set up headless X 120 windows environments on servers or use specialized CI services that produce 121 screencasts of test runs. 122 123 Testing fully-server-rendered applications, on the other hand, is amazing. A 124 vanilla testing setup with Rails & [25]RSpec can give you fast, stable, 125 concise, and actually-useful test coverage. You can actually assert for 126 behavior and navigate through an application like a user would. These tests are 127 solving a simpler problem - making requests and parsing responses, without the 128 need for a full browser or headless browser, without multiple kinds of state to 129 track. 130 131 Not only do the tests work better, the testing culture is a completely 132 different universe. There are entire books written about how to write RSpec 133 tests that catch bugs, allow software evolution, and aren’t filled with 134 boilerplate. 135 136 Gems are so powerful 137 138 Powerful and dangerous. 139 140 I’m used to modules as they work in other systems - Python, Node, Elm, and so 141 on. They provide objects, functions, and variables that you can import and 142 combine into your code explicitly. Usually they sit on some specific level of 143 abstraction - it’s a utility for connecting to servers or a React component you 144 can use. 145 146 Gems can do so much more. You install something like [26]Devise into your 147 system and it adds views, routes, methods, utilities, you name it. It’s not 148 like “loading some functions”, it’s more like composing a whole different app 149 into your app, implicitly. 150 151 This is obviously terrifying. It means that you can’t look at your directories 152 of views and your file of routes.rb and know what exists at a glance. There are 153 other layers, lurking in the ephemeral space of third-party code. They interact 154 in serious but uncertain ways. 155 156 But it’s also pretty incredible - the idea that something like [27]passport, 157 Node’s middleware, could instead be a full-fledged authentication system. It 158 means that you have to write a lot less code, and it also means that the people 159 who use that code have a lot more code in common. That gems can work on a 160 higher level of abstraction, making it possible to cobble together software 161 faster, to write less ‘glue code.’ 162 163 There’s so much good writing about Rails 164 165 Even if you don’t write Ruby, you should pay attention to [28]Sandi Metz. She’s 166 incredibly wise and has so many incredible ideas to share. 167 168 And then there’s [29]arkency, [30]ThoughtBot, and so many other thoughtful 169 writers with years of experience in Rails. Sometimes it’s a little shocking to 170 google for some obscure problem and see a decade of discussion about it. 171 172 The best practices are also formalized into tools like [31]Code Climate and 173 [32]reek. I’ve never seen so many actually-useful suggestions come out of 174 automated systems as I did in the world of Ruby and Rails. 175 176 Ruby 177 178 Ruby is a pretty pleasant language to work in. Sure, it has a lot of syntax and 179 a sprawling standard library, but you don’t have to use all of that if you 180 don’t want to. It took me a while to adjust to the object-oriented way of doing 181 things - in particular, the idea that you can’t just have a free-range function 182 floating out there, unassociated with a class or module, like you can in 183 JavaScript. And you can’t just create an arbitrary one-off object - you either 184 need to define a class to create an object, or use a Hash to store data. 185 186 But Ruby’s standard library isn’t that huge. I’ve seen JavaScript’s ‘standard 187 library’ grow a lot too, and frankly it’s nice to have methods like [33] 188 String.prototype.padStart instead of having every little thing in userspace. 189 The only part that felt actively weird was [34]activesupport - a gem that 190 extends Ruby’s core objects, but is part of Rails. It felt weird to have string 191 methods that would only work if your environment was Rails. 192 193 The [35]Dash app for documentation rocketed from my pile of unused tools to an 194 absolute must-have. In the world of Ruby and Rails, with most gems having 195 pretty good, semi-standard documentation, you can search for, and get answers, 196 super fast. The Ruby language documentation and the Rails documentation is 197 absolutely great. The JavaScript equivalent - [36]MDN - pales in comparison. 198 199 The bad 200 201 The asset pipeline 202 203 Remember SASS and the YUI Compressor? These are, unfortunately, defaults in the 204 [37]asset pipeline. There’s [38]Webpacker too, which has a parallel approach to 205 CSS and images as the asset pipeline. It has [39]opinionated integrations with 206 stuff like React. Ah, and I should mention that Rails’s [40]JavaScript 207 utilities are written in… CoffeeScript. 208 209 I get it - it’s hard to keep up with the latest trends in frontend. But this is 210 one area where Rails’s strong backwards compatibility feels iffy. I wish that 211 Rails was more opinionated about the frontend, and that it had better opinions. 212 213 Best practice churn 214 215 In Smalltalk, everything happens somewhere else. - [41]Adele Goldberg 216 217 Ruby, as today’s Smalltalk, has the same issue. The community venerates small - 218 that methods should be short, files should be small, complexity should be 219 controlled. This begs the question of where it all goes - certainly not in 220 controllers, which should be skinny, and not in views, which should have very 221 little logic at all, and maybe [42]not in models either. Maybe in [43]Service 222 Objects, or policies, or decorators? 223 224 I found myself falling victim to this. I’d try to win CodeClimate’s approval by 225 moving code around, perfecting the art of making everything small or at most 226 medium-sized, extracting concerns until most files looked okay. This was time 227 well-spent on learning, but I have to admit that it doesn’t actually matter for 228 an early-stage startup’s product. 229 230 In stark contrast to the folks who say that Rails is for prototypes, there’s a 231 lot of attention paid to long-lived engineering efforts - adopting patterns 232 that let many team work on the same ‘monolith’, identifying [44]shotgun surgery 233 - a term I first heard from Sandi Metz. 234 235 ActiveRecord is great, except when it isn’t 236 237 One of the hardest bugs we encountered happened with ActiveRecord. We were 238 creating a set of changes to apply to a model, using their in-memory instances 239 to do some stuff, and then finally applying them. This broke because one of the 240 ActiveRecord methods automatically ‘committed’ those changes, quietly. 241 242 ActiveRecord is kind of like this - a lot of the times it’s pleasantly 243 implicit, letting you just assign a value and automatically saving that to the 244 database. But then it’ll do something implicitly that you don’t want to happen, 245 and figuring out why this happened and how to stop it from happening is a real 246 challenge. 247 248 Most of the time, to be clear - it’s a really great system. It provides lots of 249 ways to generate efficient-enough queries, knowing full well that SQL 250 performance is often the bottleneck of web applications. Most of the time it’s 251 really nice that it automatically casts and deserializes query results. But 252 when it goes bad, the diagnosis and the cure can be pretty ugly. 253 254 The other issue with ActiveRecord is that it has efficient methods and 255 inefficient methods right next to each other, because it automatically turns 256 your ‘query builder’ into an array when you call array-like methods. So, for 257 example: 258 259 Dogs.all.max_by(&:height) 260 261 Is wildly inefficient. It might fetch and deserialized a million records just 262 to sort them and give you the first. On the other hand, 263 264 Dogs.order(height: :desc).first 265 266 Is fast - it sorts in the database and fetches a single record. Rails is both 267 offering smart and easy ways to write optimized code, but also making it really 268 easy to write inefficient code. 269 270 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 271 272 A Rails-like framework is a really good thing to have in your toolbox, and 273 there’s a lot to learn from the Ruby community. My hope is that we see these 274 sorts of abstractions in new languages and frameworks, and see more of the Ruby 275 community’s culture filter into the programming world. 276 277 February 18, 2021 [45]Tom MacWright ([46]@tmcw, [47]@[email protected]) 278 279 280 References: 281 282 [1] https://macwright.com/ 283 [2] https://macwright.com/reading/ 284 [3] https://macwright.com/photos/ 285 [4] https://macwright.com/projects/ 286 [5] https://macwright.com/drawings/ 287 [6] https://macwright.com/micro/ 288 [7] https://macwright.com/about/ 289 [8] https://rubyonrails.org/ 290 [9] https://nextjs.org/ 291 [10] https://www.rust-lang.org/ 292 [11] https://macwright.com/2020/05/10/spa-fatigue 293 [12] http://boringtechnology.club/ 294 [13] https://macwright.com/2020/10/28/if-not-spas 295 [14] https://twitter.com/tmcw/status/1321133460501585922 296 [15] https://developers.google.com/web/updates/2015/05/automatically-pause-on-any-exception 297 [16] https://www.html5rocks.com/en/tutorials/developertools/sourcemaps/ 298 [17] https://github.com/facebook/regenerator 299 [18] https://github.com/deivid-rodriguez/byebug 300 [19] https://github.com/fxn/zeitwerk 301 [20] https://weblog.rubyonrails.org/2005/8/25/10-reasons-rails-does-pluralization/ 302 [21] https://nightwatchjs.org/ 303 [22] https://jestjs.io/ 304 [23] https://enzymejs.github.io/enzyme/ 305 [24] https://www.cypress.io/ 306 [25] https://rspec.info/ 307 [26] https://github.com/heartcombo/devise 308 [27] http://www.passportjs.org/ 309 [28] https://sandimetz.com/ 310 [29] https://blog.arkency.com/ 311 [30] https://thoughtbot.com/blog/ 312 [31] https://codeclimate.com/ 313 [32] https://github.com/troessner/reek 314 [33] https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/String/padStart 315 [34] https://web.archive.org/web/https://rubygems.org/gems/activesupport/versions/6.1.1 316 [35] https://kapeli.com/dash 317 [36] https://developer.mozilla.org/en-US/ 318 [37] https://guides.rubyonrails.org/asset_pipeline.html 319 [38] https://edgeguides.rubyonrails.org/webpacker.html 320 [39] https://github.com/rails/webpacker#integrations 321 [40] https://github.com/rails/rails/tree/main/actionview/app/assets/javascripts 322 [41] https://en.wikipedia.org/wiki/Adele_Goldberg_%28computer_scientist%29 323 [42] https://thoughtbot.com/blog/skinny-controllers-skinny-models 324 [43] https://codeclimate.com/blog/7-ways-to-decompose-fat-activerecord-models/ 325 [44] https://en.wikipedia.org/wiki/Shotgun_surgery 326 [45] https://macwright.com/about/ 327 [46] https://twitter.com/intent/follow?screen_name=tmcw&user_id=1458271 328 [47] https://mastodon.social/@tmcw