by bibyte on 8/7/19, 7:10 PM with 711 comments
by rossdavidh on 8/7/19, 9:39 PM
There was nothing, technologically or organizationally, preventing web development from becoming so complicated, such that websites whose desired functionality would be satisfied by static html are now several meg of thousands of lines of javascript on top of a few dozen included libraries. There was nothing preventing that. Therefore, it happened.
Developers don't really feel like pointing out, "you should get rid of me and just pay for someone to do html". Managers don't really feel like pointing out "the project I'm managing doesn't really deserve a team of developers, you should take them away from me." The designers don't feel like pointing out "all we really need is to put this well-formatted text on a page with the company logo on the top and one or two image files, and that will communicate the information just fine, you can get rid of me." The even-higher-ups don't really feel like pointing out "I don't need to hire all these people, you can take this VC money and return it to the original investors". The VC's don't feel like pointing out, "there's not really a use for this much money in the tech world right now, so you can just stick it in a bank account and wait for a real need."
So, it doesn't happen, because there was nothing to prevent it.
by jjcm on 8/7/19, 8:25 PM
Thus, growing pains and overly complicated frameworks/plugins/other bits slapped together to address the core issues which ballooned the complexity. But we all knew that.
What's more impressive is that things are getting less complicated. Browser monoculture, while bad for the freedom of the web, is going to simplify things radically. Only having to target rolling releases for Firefox and Webkit? That's a dream from a cross-plat support side. Also javascript is actually pretty good now. Webcomponents once Chrome 77 drops and Firefox matches parity will finally be production worthy, and they're actually quite easy to build. I've been making static sites with them recently and it feels so much cleaner than going with a React or other component library approach that uses a vdom. The HTML is readable again, and there isn't any needed compilation step or crazy npm dependencies.
While it will never be as easy as it was back in the mosaic days, web dev is simplifying.
by reaperducer on 8/7/19, 8:04 PM
A few weeks ago I launched a new web site for a health care company. HTML, PHP, CSS, and MySQL. No frameworks. No javascript. No garbage.
It replaced a site that was a mess of javascript libraries on top of frameworks on top of a half dozen jquery version and on and on and on.
The company administrators, doctors, practitioners, etc... are so happy with the new no-nonsense site that they raved to the parent company about it. The parent company's IT department looked into the site, asked me some questions about it, and yesterday I was been asked to do the same type of cruft-free rebuild of three other sites that the company owns.
Not buying into the framework-of-the-day hype just landed me job security for the next two years.
by neya on 8/7/19, 8:10 PM
One of my clients was an airline. He asked if we could track the usage stats of his in-flight entertainment system (just stats, no PII). I thought about it and said yes. Even I was amazed I said because a decade ago it wouldn't be possible. I got it done with service workers (because flight wi-fi isn't a 100% always on connection).
It was a boon for me because I made a lot of money on that project and I needed the money. It's also a curse because every browser has all these invisible code running behind the scenes whether you want it to or not.
I miss those days where everything was just a bunch of tables and default system font-powered webpages with plain blue links. No CSS, no popups, just html and pure content.
by shados on 8/7/19, 9:43 PM
Basically, developers can support complexity <x>. No matter what you do to simplify it, you'll always push toward <x>, just getting more value out in the process. If you don't, your competitors will.
Also, even the most complex rollup/webpack/react/redux/styled-component/npm/blah/blah/blah setup has nothing on the kubernetes/kafka/databases/queues/whatever infrastructures of the world.
It just turns out that while your little pet project that cuts every corners to get something working has little on software development that requires everything to line up, be fast, reliable, comply with all the laws and regulations, and scale once your company grows and you suddenly have hundreds or thousands of people that need to work on the same stuff.
One thing to add: some people actually like setting stuff and optimizing infrastructure, build processes, setup, and pushing it to its limits. So development stacks quickly get optimized for a very common way to staff organizations: have a team that handles all the setup and infra, and a bunch of teams that build on top. So when you try to have 1 person or 1 team do everything, it feels like too much.
by goatinaboat on 8/7/19, 8:16 PM
Someone else mentioned SPAs. I 100% guarantee that no user in the world ever wished for their “back” button not to work as they expected.
Calling themselves “full stack engineers” is another symptom of this.
by tschellenbach on 8/7/19, 7:50 PM
by chiefalchemist on 8/7/19, 8:55 PM
Nah. Nobody cares. I think that bears repeating...Nobody cares.
My mum has __never__ said to me "I love React" or "Isn't Bootstrap the best." But she __has__ complained to me about a site/app being buggy, confusing, difficult, etc. I've said the same to myself too many times to count. I think we all have.
A couple months ago, I read an article about (frontend) tools and how important X, Y and/or Z are. The author's product had a repo on GH. There were 700+ issues. All I could think was, "Apparently those tools have given his team a false sense of security. Looks like they've bitten over more than they can chew." I filed the article under: Just because you can, doesn't mean you should.
Years ago, I remember reading an answer on Quora (when Quora still mattered) that included:
"A fool with a tool is still a fool."
I don't think I'll even forget that line. It's so true. I'm not against tools. I'm just not in favor of those who claim tool X, Y or Z is a panacea.
by daotoad on 8/7/19, 11:42 PM
If you've ever written or worked on a classical desktop application, you'll know what I mean. Since (at least) the 90s there have been visual UI builders and geometry management tools that allowed you to easily define a responsive user interface and connect widget events with calls to back end code. With the advent of CSS Grid and Flexbox, we finally have two decent geometry management tools--it only took 20 years. Except I can't use them yet, I have to support terrible old browsers (like IE7) that are embedded in customer applications.
We've been involved in reinventing a pretty, zero-config version of X-Windows, and we're only about 60% of the way done. How much further would we be if we didn't insist on building it in a hopped up word-processor?
by vcanales on 8/7/19, 8:24 PM
I've been working on the server- and client-side of the web for about 10 years now, and I can't recall running into these "wandering pixel" issues they talk about in the past ~4 years.
Frontend development is not just about building a cute website for grandma anymore, and that's why it's complicated. We're passing more responsibilities to the browser, and we have to care about performance when doing work there, and that's why it's complicated. Writing performant software for an unknown machine is hard, and that's why it's complicated.
by mark_and_sweep on 8/7/19, 8:06 PM
Recommended reading:
- "Responsible JavaScript: Part I" by Jeremy Wagner (https://alistapart.com/article/responsible-javascript-part-1...)
- "A JavaScript-Free Frontend" by Matt Reyer (https://dev.to/winduptoy/a-javascript-free-frontend-2d3e)
Disclaimer: I'm from Germany, but have a 6-y/o Android phone and live in a rural area with poor EDGE connectivity (~50kb/s). Most "modern" websites don't load/work for me.
by Yaa101 on 8/7/19, 8:55 PM
More complicated answer: The weak side of the open source world is that it attracts a lot of developers with big egos who think that their fancy solution to a situation without problems is the best one. And they all compete with their too big egos and complicate our open source world.
Real answer: too much memory, this causes people to come up with too complex systems as there is hardly a limit to their dreams, real inspiration comes from dealing with limits. That is why programs from yesteryear are often so good when looking at what they do within their constraints.
by skrowl on 8/7/19, 7:57 PM
by heyoni on 8/7/19, 7:31 PM
by aphextron on 8/7/19, 9:35 PM
by earnubs on 8/7/19, 8:03 PM
So now you can be more expressive, target more browsers easily, have greatly simplified script loading and module management, much, much better debugging tools, with less effort.
The tools make it easier for you to be a good developer.
Good web development was never easy.
by Razengan on 8/8/19, 12:36 AM
"Modern web development" tries to reinvent too many wheels.
I wish more developers/directors would try to approach this from the other direction: Making native OS apps as easily discoverable, installable, accessible, navigable, sharable, restorable, updatable and uninstallable, as websites:
• I want to be able to type "photo editor" and see Pixelmator on macOS, Paint.NET on Windows, Photoshop on both, and GIMP on Linux.
• When I choose an app it should open as fast as a website, loading incrementally, but remain available even after I'm offline (except for internet-dependent features) like other native apps.
• It should always launch the latest version, but should also allow me to "freeze" and locally archive a specific version.
• It should adopt the look, feel, and all the special features of the OS and device I'm currently using.
• I should be able to capture the UI state as a "link" and share it with other people, for example a specific page in the preferences window, or a particular workspace layout.
• It should remember my preferences across machines and OSes.
• After I have stopped using it for a while it should not take up any local storage on my machine.
We already have high-performant, energy-efficient UI toolkits and graphics and audio engines, we already have task switchers and various accessibility features, and we already have everything else that every web app is constantly trying to reinvent, each in its own half-baked way.
It's a constant shuffle of N steps forward, N+m steps back. "Oh look! A cool new technology for native apps! But wait, let's take a very roundabout detour to reimplement this in a crippled imitation so that browsers can host it too."
by pkalinowski on 8/8/19, 10:13 AM
I'm too stupid for Electron, though. I wasted many hours trying to debug weird errors which I never know if are caused by Electron itself, bundlers, npm modules not compatible with it or anything else like division between main/renderer process. I just can't grasp it.
I ended up with separate node.js server which takes care of communicating with OS and standard web app with the simplest bundle setup I have found (because I don't understand webpack config, too): Parceljs. Frontend and backend exchange information using socket.io with copy-paste examples from their website.
I couldn't believe it CAN be so simple! Maybe it's not "true" desktop app (it just opens in user browser), but works exactly like I want.
by MadWombat on 8/7/19, 10:17 PM
So now we are using HTML, CSS, jQuery, babel, webpack, webpack dev server, webpack hot reload plugin, webpack plugins to load resources and a minifier. But hey, keep it simple, right?
by dwheeler on 8/7/19, 8:48 PM
Everything is a trade-off. The usual cause of excessive complexity is engineering based on the latest fad, instead of what is actually the best choice.
by gedy on 8/7/19, 7:57 PM
by skc on 8/7/19, 8:42 PM
When he was born I used to dream of having him look at a button on a web page and him asking me "what actually happens when I click that button"
I shudder to think where I would even begin to answer that today.
by maximp on 8/7/19, 7:57 PM
The insane number of tools and processes was the motivator behind Hackterms (https://www.hackterms.com) - the Urban Dictionary for programming terms, and a sort or modern Jargon File. We're now at ~1200 definitions and I reference it regularly.
by ht85 on 8/7/19, 8:37 PM
To many people, backend and infrastructure work is sexy. They would happily get their hands dirty and spend insane amounts of time debugging mind-boggling issues as long as they happen server-side.
Frontend work on the other hand is not, it seems. Maybe that's because the biggest problem of high quality frontend development is to figure out how to deal with users (who most deveopers seem to abhor), and how to navigate the, say... complicated circumstances that are browsers.
Maybe building something that has to work in an environment you can't control is just a little too stressful for most devs?
Personally I really don't care. I love doing great architectural work, but in the end I'm all about expertise and I don't discriminate. If I have to step-debug inside webpack plugins or understand Safari CORS handling quirks, so be it. I'm a software engineer, but to me sexiness is in relentlessly looking for an edge, happy customers and signed contracts.
by jandrese on 8/7/19, 7:46 PM
The generators can also do things in really dumb ways that make it slow or resource intensive and you'll have no clue why your elegant solution is running like a dog unless you go through and read the horrible generated code to see what it is actually doing.
by chopdcheez212 on 8/8/19, 2:17 AM
JavaScript I was paid to write in 1995 had a shelf life of 6 months given Netscape’s release schedule and I had to compete on performance with Delphi 3 native compiled code.
If you think web development is too complicated these days, you should just dump it all and write cgi scripts in Perl again.
Or if you’re an academic, Smalltalk + Seaside.
If there is a problem with modern web development, it is the curse of too much choice. Focus on your end goal, and the choices available will become a blessing, not a curse.
by president on 8/7/19, 8:23 PM
by buboard on 8/8/19, 12:03 AM
Was it the inflow of too many mediocre graduates that needed help preventing themselves from writing buggy programs? The social credits that people added to their CV for making yet another uber-popular javascript thingy that got a lot of github stars? Or was it bored engineers in the major web companies who have nothing to do because their advertising money cow is already being milked, so they decided to build toys to sabotage everyone else?
by cutler on 8/8/19, 2:12 AM
by johnwheeler on 8/7/19, 7:47 PM
by iamsb on 8/8/19, 6:43 AM
I recently went to work a two big company, which is considered highly agile and progressive. I was tasked to build a pricing engine to customize prices for a static inventory of products. Number of products were about 30 and did not change month to month. The architecture suggested we use close to 18 different technologies, including caching systems, message bus, reactive programming. All to give 10 possible prices for 30 products. So essentially 300 values and a system to change the rules which determined the prices based on 3/4 parameters. I built the entire system in 2 days which was ready to be tested, but was told to scrap the work as it did not follow the internal development guidelines. Not to mention I was reprimanded for not including team and calling a complex problem trivial as it increased signaling risk internally. Apparently they are still building that system with 5 engineers, 2 analysts, 1 devops person, and 1 team lead. I left the company 2 years ago. And interestingly each and every person on the team got a promotion, multiple performance bonuses in last 2 years.
by bitwize on 8/7/19, 8:13 PM
Also, an application software vendor had A(m,n) problems and thought "What we need is to make the Web into a universal platform capable of supporting any application." Now they, and the world, had A(m+1, n+1) problems.
by mirimir on 8/8/19, 12:22 AM
Although lots of cool stuff is possible, I suspect that very often, it's about profiling and tracking users, and not about delivering content. That's certainly my experience as a user. I routinely browse sites that call resources from numerous sites, and want to run multiple scripts. And all they're ostensibly doing is displaying some text and images.
by bmniotjoti on 8/7/19, 11:36 PM
We as a civilization haven't really had a need to reconcile this I think on the scale that we do today. Nothing was "out in the open" so much so as it is today. But now that I can see it somehow my expectation is that I should immediately understand it.
As more and more information becomes available to the masses we all need to start recognizing that just because we can see it doesn't mean we should expect to be able to understand it.
Real knowledge about how to solve complicated problems takes time and discipline to get to the place we want to be individually and collectively.
by neop1x on 8/8/19, 11:08 AM
by drej on 8/8/19, 10:25 AM
Given how much I spent doing things this way, I'm quite reluctant to building JavaScript, let alone CSS (though preprocessing first came to CSS, if my mind serves me right). I find the write-test-update loop to be much tighter for frontend work, so I'm quite fond of the old ways of doing things (and all my side projects look the same way as they did back in the early 2000s), but I understand that modern build tools are getting more and more performant. Not as performant as "Save" though :-)
by ummonk on 8/7/19, 8:14 PM
by prince005 on 8/8/19, 6:52 AM
by jonnycoder on 8/7/19, 9:47 PM
I'm developing a dashboard, nothing fancy but nothing simple. I am also reasonable and think one can build off of all the well established UI components out there such as Vuetify. I now have a lot of free time to spend on the backend API, which is what I happen to enjoy doing.
As a fan of end to end or integration testing, I could care less about test driven development on the front-end.
by dnprock on 8/7/19, 8:09 PM
I still stick with Ruby on Rails stack. It has a bunch of good practices and patterns built up over the years.
On the good side, I set up a static Jekyll site using Github Pages. It comes with SSL from letsencrypt. I got to run my website for free without much setup. I assume I don't have to worry about scaling it either. No server, no nginx, no SSL config. It's amazing!
by _hardwaregeek on 8/7/19, 10:37 PM
Yes, web development is complicated. Yes, that complexity is at times completely unwarranted. But, that does not mean that the originators of this complexity were simply being stupid. A company like Facebook has a simple requirement: keep users on the site. If that means building a completely overengineered platform or inventing a totally new front end framework, sure, go for it! As long as it keeps the experience so smooth, so seductive that people scroll for hours, absorbing ads and producing money for Facebook.
I wouldn't be surprised if someone did some user testing and the SPA with no jarring refreshes and fluid page changes scored significantly higher on user engagement.
If you look at the companies that have really fancy JS libraries, they're generally places that rely on people using their apps continuously. Netflix? They're practically the poster child for internet addiction. Twitter? Yep. Even old school media companies like the New York Times are starting to use fancy web technologies. Clicks and eyeball time is money. Stuff like React and transpiling and webpack are small prices to pay if they keep your front end developers happy and your website nice and shiny for your average user.
by exabrial on 8/8/19, 12:27 AM
by hacker_9 on 8/7/19, 8:25 PM
by oeijfgoierjg on 8/7/19, 11:09 PM
People seem to get openly frustrated if they can't learn overnight all there is to know about a discipline that has taken some people years and maybe even decades to master.
We as a civilization haven't really had a need to reconcile this I think on the scale that we do today. Nothing was "out in the open" so much so as it is today. But now that I can see it somehow my expectation is that I should immediately understand it.
As more and more information becomes available to the masses we all need to start recognizing that just because we can see it doesn't mean we should expect to be able to understand it.
Real knowledge about how to solve complicated problems takes time and discipline to get to the place we want to be individually and collectively.
by mholt on 8/7/19, 9:42 PM
There is a lot to be said for simple, time-proven technology.
[1]: https://medium.com/@mattholt/its-2019-and-i-still-make-websi...
by proc0 on 8/8/19, 5:38 AM
Ironically some of the tools in the article will probably grow old which is part of the difficulty here. People learn the libraries, then it seems a new, better library comes along and makes things even better. Everything comes together once you learn it's about managing the algorithms, data, as well as rendering to the screen.
by tempslkjs on 8/8/19, 7:36 AM
by arendtio on 8/7/19, 8:32 PM
But what I feel quite frustrated about, is the pace at which tools become obsolete. It seems that you can be happy if you have two projects which actually use the same set of tools. And every time you want to fix a bug in some project which you didn't touch for a few months, you run into some kind of trouble while updating the dependencies.
Just today, I spent an hour getting a project to build again, which I built two years ago. I just hope that in the near future the community will settle on a common set of tools which will become stable and stay for a decade or so.
by tempsy on 8/7/19, 8:06 PM
For personal projects, React on Rails (+ Redux) has been a very productive setup. Once you get over the learning hump of how to set up a project it's basically the same thing over and over again.
by rixed on 8/8/19, 12:17 PM
For my current project I needed a GUI that I initially planned to implement in the browser, but then I started to question this assumption that web development is simpler, and decided to try to refresh my C++ and use Qt instead.
So far it's not a disappointment.
I wrote a bit recently about that here, if you are interested and can bear poor English:
by iamleppert on 8/7/19, 11:58 PM
In any given technical environment that has sufficient time to mature there will be more information to process, more choices, and more surface area. It's just the principle of intellectual entropy. Stuff tends to get larger and more complicated and specialized because it is.
If you want something simple, why not invent it? There's nothing stopping the next person from coming along and coming up with something that gets the same job done a lot easier than what we have now.
Of course, the first time you try to implement a tiny portion of a browser primitive like drawing a table you'll quickly see why things are complex.
by chrshawkes on 8/7/19, 8:52 PM
We can complain about it all we want but if you drop out of this for a few years, good luck. I know tons of senior dev's who have no idea what modern tools to use. They don't understand modules, loaders, config files.
Trust me on this, the front-end guys like myself, hate the backend guys that come along and criticize everything they know nothing about. They also get in the way as they bumble and stumble over concepts UI guys already know. There is a huge distinction these days with backend and frontend web development. Each should avoid criticizing the other.
by mister_hn on 8/8/19, 8:40 AM
To all the people saying that you want a native-app look, then go for the native way. The web-apps will _NEVER_ be as fast as the native ones, just because, the native ones will be faster and faster as the hardware specs are going up. I personally hate to go through web-development, even if it's a desktop client: stop wrapping around web views.
by amiga-workbench on 8/7/19, 7:48 PM
So now we have meme frameworks and transpilers and hysterical build toolchains all for a platform for sharing documents.
by winrid on 8/8/19, 6:16 AM
Also winricklabs.com
It's a nice breather after writing AngularJS for six years... I'm sure with the next complicated UI I'll use some big framework (Spring+Leaflet or Angular...)
However, I think the goal should be to reduce complexity and have your business/product be focused as possible.
Watch.ly is literally just live hit counters. The live part gets challenging at some scale. But the product is focused.
by hhas01 on 8/8/19, 9:15 AM
Given the choice between learning the business – requiring lots of talking to users to understand what they do and why they do it, and then writing very plain uninteresting code that servers those business users’ needs – or learning whizzy new CV-enhancing technologies and using them to invent excitingly complicated and interesting software problems so they can spend all their time solving those problem instead, which do you think the indolent bums are going to embrace?
by madrox on 8/7/19, 10:32 PM
by wildduck on 8/8/19, 12:36 AM
Rise of the SPA is great. However, I think unless there is a good reason for having a SPA such as hybrid mobile app development, then there is no point of using SPA.
Normal server side web page rendering with latest modern version of jQueryish/Zapeto should be just fine for the non app type of web development.
by quickthrower2 on 8/7/19, 9:53 PM
* Javascript has evolved from simple requirements e.g. flashy text to a full blown application environment, but there is relatively little base functionality included (e.g. compared to what you get in Java or C#) so people need to write those libs separately.
* The separate libraries above need to be included in your code, and copy/paste is not maintainable so some kind of module system is needed but JS doesn't come with one, so various people rolled their own and they competed for mindshare. 10 different standards.
* As above for UI frameworks. JS only has a basic framework included: the DOM. It's pretty good but doesn't scale "naked" to more complex apps, so again the community creates a whole bunch of those.
* As above for application frameworks (UX frameworks)? such as routers so that you can press the back button.
* Ah the back button - an example of where apps on the web have to care about something that desktop apps don't. There are other examples. Mixing http/https. Mobile responsive.
* CSS - I imagine originally designed for "documents" - now has to work well with "web apps".
* Javascript again - no static typing, many devs find it too easy to make mistakes, we have Typescript and other compile to JS languages.
* Because of the above complexity with have Babel, Browserify, Webpack etc. (one is never enough!)
* Then some genius decides to use JS for the back end and we have Node, NPM, and then because Node is so useful it becomes the front-end tool chain provider. So you use NPM to install Webpack, and the newbie has to figure out difference between dev dependencies and ordinary ones etc.
* git clone https://github.com/... and who knows what stack they've used. Oh they are using Browserify. Never used that. 2 hours reading the docs.
* Each framework then finds ways to get more complex organically. React, Redux etc. and all that.
* Again every other programmer is using 10^10 different combinations of frameworks and techs. Compare to C# Desktop development where it might be they are using Winforms or WPF - that's your only problem!
I could go on. I still copy and paste my favorite templates for getting started with NPM/Webpack/Typescript. Not sure I could demo how to get there from scratch again. All the recommended ones online seem to have beartraps in them and stuff I don't understand.
But remember you all you need is a <script></script>! It's easy!
by kareninoverseas on 8/8/19, 2:46 AM
by dandersh on 8/7/19, 8:13 PM
The current complexity is a result of addressing these issues and doing so at a rapid pace.
by brillout on 8/8/19, 9:58 PM
- React or Vue?
- Redux or stateful components?
- GraphQL or REST or RPC?
- SSR or SPA?
- ...
We are missing a framework with sensible defaults for all these questions. And that does a good job at explaining what decisions one should take.
I'm trying to build such framework: Reframe (https://github.com/reframejs/reframe).
by te_chris on 8/7/19, 8:25 PM
by Rothnargoth on 8/7/19, 8:14 PM
The software stack is a natural phenomenon that adds to web design's complexity. Ever changing and never complacent, it continues to shape new features that weren't conceivable in the past.
by mbrodersen on 8/9/19, 11:52 AM
by Double_a_92 on 8/7/19, 8:00 PM
by xhruso00 on 8/8/19, 6:09 AM
by AlexDragusin on 8/7/19, 9:25 PM
by justicezyx on 8/7/19, 7:47 PM
Setup a web server: * VM from Cloud provider * Installing a bunch of staff * Throwing in a k8s cluster * Service mesh * Yaml files * Various network configuration * DNS * Firewall * Security etc.
by malyk on 8/8/19, 3:19 PM
There are other reasons, but that is a huge one.
by Ididntdothis on 8/7/19, 9:44 PM
by gigatexal on 8/8/19, 1:25 AM
I found backend development much saner to reason about.
by fishingisfun on 8/8/19, 3:03 AM
by username90 on 8/8/19, 7:48 AM
by randomsearch on 8/7/19, 11:04 PM
by sylens on 8/7/19, 9:56 PM
by ggregoire on 8/8/19, 1:18 AM
by janpot on 8/7/19, 9:37 PM
by segmondy on 8/7/19, 10:02 PM
by ProAm on 8/7/19, 8:13 PM
You simply cannot be a monolithic craftsman, who wields only a few tools well. Because if you do that, in 10 years you'll be left in the dust complaining about ageism vs knowing new tooling
I'm not advocating this is a good thing, but its the environment we have fostered for ourselves and now we don't like it.
by edem on 8/8/19, 7:33 AM
by wintorez on 8/7/19, 8:33 PM
by smashthepants on 8/7/19, 8:49 PM
by commandlinefan on 8/7/19, 8:37 PM
All the tools that are meant to simplify it.
by jonplackett on 8/7/19, 8:24 PM
I get that the flash plugin sucked and was a battery hog.
But the development environment was quite decent. I'd love to be able to make HTML5 content in the same kind of way.
by moocowtruck on 8/7/19, 7:44 PM
by innocentoldguy on 8/8/19, 4:45 AM
by tomc1985 on 8/7/19, 9:41 PM
by rb808 on 8/7/19, 8:27 PM
by loblollyboy on 8/7/19, 8:36 PM
easy
by butterisgood on 8/7/19, 10:00 PM
by austincheney on 8/8/19, 2:57 AM
This is a learning process. It isn't something a tool can magically do for you. If you think you can nail this in two hours or by dicking around with some tool you are lying to yourself. Don't worry, everybody with half a brain will see you for exactly what you are, so be honest with yourself that you probably aren't some software rockstar and learn to take the primitive dumb stuff very seriously.
If you are hoping a tool will magically do it for you then you will continue to suck. The web will continue to be a complicated mystery. You will probably continue to cry about how hard life is and people won't want to be around you.
At the most primitive the web is HTTP and HTML. Don't over think this. If you understand those you have the web. You don't even need URI. URI is an added convenience.
Next, build less shitty HTML. Any HTML no matter how ugly, inaccessible, presentation-oriented, user-agent centered, and otherwise broken is still a good start so long as it displays content to a web browser. The first goal of HTML is something that just works... somewhat. Improve on this. Learn to write better HTML and better content. Most people don't take HTML seriously, but this is really the most important part of building a good web experience.
Third, learn some presentation using CSS. This will take some practice. This takes some real practice. It isn't hard, and you can do really almost anything with CSS. Even still people FEAR the CSS.
Fourth, be a damn programmer and stop being afraid to read code. Learn to write JavaScript. Learn the concept of scope. Learn to really read code. A bunch of tooling and test automation isn't going to read the code for you, though many developers seem to think that is exactly what layers of unnecessary tools are for. Keep this very simple at first. Learn some basic events and then bail out. Don't immediately think you are some kind of architectural wizard, because you will fail. If you think a framework will solve this for you then continue to be mystified and confused.
Fifth, learn the DOM. The DOM is the heartbeat, or assembly language, of the client-side application experience of the web. The DOM is what binds HTML to the browser and CSS/JavaScript to HTML. Without the DOM you have a plain text experience with really bad presentation that you shouldn't be using. Frameworks won't teach you this. The best way to learn the DOM, how it really works, is to learn XML Schema. I strongly suggest this even if you never use XML or schemas ever again.
Build on that. Baby steps. It takes some effort but it isn't complicated.
by apatheticonion on 8/8/19, 12:36 AM
A back end server with this configuration would have a handler layer for IO, business logic layer and data layer.
DI and IoC ensures that the IO layer doesn't know about the data layer as the logic layer takes it as a dependency. The IO layer consumes the logic layer and the logic layer doesn't care/know about that.
Front end is exactly like this. The data source is REST rather than a database. The UI layer is an event source and consumer. The logic layer notifies its consumers (the UI layer) that there has been a change. The rendering library doesn't matter, it's just a mechanism to represent the events it consumes.
The problem in front end is there has been no rational mechanism for DI in the libraries. React relied on the service locator pattern to deliver dependencies so to ensure quality decoupling you would need to have wrapped your components with a million functions.
``` withTheme(withUsers(withSomethingElse(withAnotherThing(Component)) ```
So rather than that, just use a single "service", redux, and forget about DI. Everything can do anything and all your logic is done inside that one thing.
Angular is no better. While they have a nice DI solution, it still relies on concrete implementations rather than abstractions. So your business logic must live within "angular services" and can not simply be "javascript" consumed by a rendering library.
Vue just missed the point entirely.
React context packaged this problem in an ergonomic box and, in my opinion, is a fantastic solution given the constraints. However at this point, the meta in front end programming is so irrational that it boggles my mind.
In the following example, you can see that I have an application which uses React as a renderer. React consumes a package from /platform. The package has no relationship to the presentational layer.
I could run the units in the package inside node or the browser console and it wouldn't care about its rendering library. The "PostStore" class exposes an API which consists of a getter and a stream. React consumes the stream, because it needs to be notified on changes.
https://stackblitz.com/edit/react-ts-biohay?file=gui%2Fapp.t...
Conversely, this loose coupling means that the view renderer is a small portion of the application. I could swap it out for another renderer. Preact, Svelte, whatever.
Front end is a mess.
by growlist on 8/8/19, 12:17 PM
by crimsonalucard on 8/7/19, 11:36 PM
If there was a way to quantify a good design vs a bad design through an algorithm then people can iterate in ways that properly and categorically makes things better rather than in ways that are opinionated, vague and possibly even in the wrong direction.
by m0zg on 8/7/19, 9:41 PM
by marknadal on 8/7/19, 7:58 PM
Modern web development is not about making things more efficient or easier to do.
It is about proving you have a job title or knowledge that makes you more hire-able, raise-worthy, title-superior than Junior Devs.
Web dev is actually easy if you want it to be, avoid the hype and use simple tools that work out of the box.
by chovy on 8/7/19, 9:36 PM
by foobar_ on 8/7/19, 9:41 PM
by s_y_n_t_a_x on 8/7/19, 7:49 PM
It's not just web development that is hard. You can't honestly say any toolchain on the desktop side is any easier.
That said, web development is getting easier now that the dust is settling and only a few major frameworks are used.
Get started with create-react-app. React is the industry standard, use it.
1) Install NodeJS
2) npx create-react-app demo