AngularJS

» see angularjs.org as it lived, on the wayback machine

~ the obit ~

i wrote angularjs in two projects around 2014 and 2015. one was a small admin tool for an internal data pipeline, the other was a single page app sitting in front of a c++ backend. the framework was already the default choice for anyone building a single page app who did not want to write their own router, their own data binding layer, or their own form validation logic. it was what you reached for and what every job posting for a junior frontend role asked about.

misko hevery wrote the first version of angularjs at brat tech llc in 2009 with adam abrons. the original use case was an internal tool for the company's pdf-to-html conversion service. hevery rewrote a six-month project in three weeks using the framework he was building, which became the founding war story repeated at every conference talk for the next five years. google brought hevery and the framework in-house, released 1.0 on october 20 2010, and used angularjs internally for the doubleclick advertising console and a long list of other products.

the framework's core ideas were two-way data binding between the html template and the controller scope, a dependency injection container for swappable services, and custom html elements called directives that let you extend the language with your own components. directives turned out to be the hardest piece of angularjs to teach and also the piece that did the heaviest lifting in real production code. by 2014 the framework had become the standard answer for the frontend half of the mean stack.

~ the rap sheet ~

~ the rise ~

the rise was steep. by mid-2013 angularjs was being used at companies like youtube and virgin america for production frontend work. the books shipped fast. ng-conf launched in salt lake city in 2014 with a thousand attendees and became the year-after-year angular community moment. tutorials on egghead and pluralsight piled up. the question of whether to use angularjs or backbone or ember was basically settled by the end of 2013 in favor of angular for greenfield projects.

the framework had real warts. the digest cycle was opaque and expensive once you had more than a few hundred watchers on a page, the documentation got better over the years but never reached great, and the directive api had four or five different configuration patterns that all did roughly the same thing in slightly different ways. people complained about the performance and the learning curve and shipped angular apps anyway, because the alternative was writing it all from scratch.

~ what it became ~

the announcement of angular 2 at ng-europe in october 2014 was the moment the framework's trajectory changed. the angular team showed the rewrite plan with a deliberately provocative slide titled "angular 1.3: the last release" that left an audience full of angularjs developers very confused. the new framework would not run existing angularjs code. the architecture was different, the templating syntax was different, the language was typescript instead of plain javascript, and the upgrade path was effectively rewrite-the-whole-app.

react had launched in mid-2013 from facebook and was picking up adoption fast. the angular 2 transition gave developers who had been considering react a clean window to switch instead of upgrading. by 2016 when angular 2.0 actually shipped, react had taken the leading position in npm install counts and in new-project starts. angular 2+ remained an enterprise-grade option at large shops but never recovered the mindshare angularjs had at its peak.

~ what came after ~

the angularjs lts program ran until december 31 2021. for the back half of the 2010s a great deal of production code was still running on angularjs because the migration cost to angular 2+ or to react was too high to justify business-side. by 2020 most large angularjs codebases had been either rewritten or wrapped behind a slow migration path. the smaller ones simply stayed on angularjs and ran on a frozen version of the framework with whatever bug reports the maintainers had time to triage.

january 1 2022 was the formal end of security patches. by 2026 the remaining angularjs production code is small, mostly internal tools at companies that have not had the cycles to migrate. the framework's larger legacy is the pattern it set for what a frontend framework should ship with out of the box. router, http client, form validation, dependency injection, build tooling. every framework since has shipped some version of that bundle, with the router and the http client as the table stakes pieces. the specific dna of two way data binding lost the argument to react's one way data flow, but the broader idea of frontend-as-application-platform was angular's contribution to the trajectory.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »