Since they didn't publish their data source, let me add a useful note: How to count the number of stars using GitHub Archive and BigQuery.<p>Naive query:<p><pre><code> #standardSQL
SELECT repo.id, ANY_VALUE(repo.name) name, COUNT(*) as num_stars
FROM `githubarchive.month.2016*`
WHERE type = "WatchEvent"
GROUP BY repo.id
ORDER BY num_stars DESC
LIMIT 1000
</code></pre>
But let's fight "star fraud". There is an easy way to register "fake" stars - if you star and unstar a project repeatedly, each time this will register as a WatchEvent on the GitHub Archive log.<p>Better query, removes duplicates:<p><pre><code> #standardSQL
SELECT repo_id, ANY_VALUE(name) name, COUNT(*) as num_stars
FROM (
SELECT repo.id repo_id, ANY_VALUE(repo.name) name, actor.id
FROM `githubarchive.month.2016*`
WHERE type = "WatchEvent"
GROUP BY repo.id, actor.id
)
GROUP BY repo_id
ORDER BY num_stars DESC
LIMIT 1000
</code></pre>
If we put all together, these are the real results:<p>* <a href="https://docs.google.com/spreadsheets/d/1aDlXrk3U1z5s0-1Is8KHOowJ_oSw79mzwAOPmK5LmeI/edit" rel="nofollow">https://docs.google.com/spreadsheets/d/1aDlXrk3U1z5s0-1Is8KH...</a><p>Projects like 'fivethirtyeight/data' lose -864 stars (23%), going down 230 places in the ranking, while projects like 'FormidableLabs/nodejs-dashboard' lose less than 1% of their stars, going up 49 places.<p>When I said 'stars fraud' I'm not presuming malice, but with these star rankings we do create an incentive :).<p>Disclaimer: I'm Felipe Hoffa, and I work for Google Cloud <a href="http://twitter.com/felipehoffa" rel="nofollow">http://twitter.com/felipehoffa</a>
I'll probably get downvoted for this, but I think counting github stars probably only reflects visibility but not actual usage. To give you an example, cf-ui is counted as one of the "best" React UI component out there in bestof.js.org, but this project is not currently meant for general consumption. This fact is reflected in npm stats.<p>I think we as a community should be very vary about these "scorecard" websites and their methodologies. These things tend to be a self-fulfilled prophecy. I'm not sure I want to see a couple of brand name companies start monopolizing our eyeballs and technical conversation. I'm also very vary about people making decisions based on these sites. While not everyone is completely different from others, I'd like people to make their own decisions based on their only thinking process rather than just jumping on the new and shiny from the big names all the time. Not that this appears to be an issue right now, just a few words of caution.
Fascinating to see Vue.js and Inferno both doing so well last year. Vue is definitely lot easier to learn than React, for new Developers, but I could be wrong since I haven't tried it a lot. Don't know how it scales up for large applications.<p>Preact is also another dark horse. This year we are going to see a few more "react-like" UI libraries.<p>In node, really happy to see Feathers catching up with other more popular frameworks like Express. First time I reached to feathers was by literally searching "Firebase Alternative".<p>AVA the Test runner, really never heard of it at all. May be I am too behind in the "Test Runners" category. I only use Mocha.<p>Also, they should have added a category for graphics libraries like Three.js, Fabric.js, Paper.js etc.
It isn't a good indication for the framework value itself.
A star in github its like "meh I heard a lot about react-thing, that look complicated but can't be that bad, I should take a look one day"
Its a self-persuading buzz
Where are the star counts from?<p>Create React App was created in 2016 and has more than 18k stars but is shown as having gained 5.6k stars in 2016 (I think it got more than that in its first week!)
Excited to see Vue doing so well on this list. I have been using it for the last few months at my day job, and I have been pleasantly surprised by it. I have tinkered with React & Angular in the past and I can second that the fact that compared to the other frameworks, getting started with Vue is a lot easier.<p>The ecosystem around it also quite mature - VueRouter & Vuex are well-tested solutions in case you have a use for it.<p>Another advantage that it has is its documentation & guides. It is quite exhaustive and easy to follow. My only gripe was that it doesn't go into the details of building a complete "single-page-app" that uses the accompanying tools (vue-cli, vuerouter, vuex etc).<p>Based on my learnings over the past few months, I have started writing a small ebook that goes over the process of building a full-fledged app.<p>I have created a small subscription form - <a href="http://eepurl.com/cvUk5D" rel="nofollow">http://eepurl.com/cvUk5D</a>. You can add your email here to get notified when I launch this book and also get access to the early release.
Really Sad to me that Ember.js gets so little love. It is hands above Angular 1 or 2 in my mind and the only JS front end framework that I feel productive in.
Hmm, Create React App is listed as "+5.6k stars", but it was launched in 2016 and has over 18,000 stars right now. Perhaps an error in the data processing?
Hello there, this is Michael Rambeau, the writer of JavaScript Rising Stars.
Thank you, everyone, for your comments.
It's very nice to see people talking about things related to my project.
As some people mentioned, after the initial release, there was an issue about the count of stars, for some projects.
I'm sorry about that, it has been fixed during the following releases.
I will try to take into account the ideas discussed there, when things calm down.
Thank you a lot!
Who says that github "stars" are a quality indicator?<p>I use stars to mark projects I am interesting in but haven't tried yet. I am sure many people do the same.
Somewhat off-topic, though it may be relevant.<p>Is it that much more difficult/complicated to have a server side rendered app with pure JavaScript sprinkled in versus SPA vs. no JavaScript?<p>The situation I'm thinking about is the following:<p>firstApp.domain.com<p>secondApp.domain.com<p>If you have an SPA then you literally can just have separate SPAs for each sub domain served statically. You could also have everything rendered server side using whatever back-end you're using. Finally, you can have server-side rendering and add JavaScript when necessary, but this seems to add complexity as your team would now need to know whatever templating language your server-side framework uses plus the front-end framework. Am I missing anything?
I don't understand why Hapi doesn't get much love these days. It's a monolithic framework, which means it's extremely opinionated in how things are done, but also it means you don't need to go hunt down 3rd party packages to cover features that every server dev needs out of the box.
I love that this made it to the report in the "React Boilerplates": <a href="https://github.com/tj/frontend-boilerplate" rel="nofollow">https://github.com/tj/frontend-boilerplate</a><p>> A boilerplate of things that mostly shouldn't exist.
The React part is especially intersting, since Inferno, Preact and React share the React-API, which basically means React blew every other framework away.
On the front end, intercooler.js[1] actually gained more stars in 2016 (~3000)[2] than Mithril (which I like, this is not to take anything away from the Mithril guys!) This was in large part due to a big HN bump in November.<p>I know it's too contrarian and idiosyncratic to get a mention on a JS survey, but I have to get he word out somehow...<p>[1] - <a href="https://github.com/LeadDyno/intercooler-js" rel="nofollow">https://github.com/LeadDyno/intercooler-js</a><p>[2] - <a href="http://www.timqian.com/star-history/#LeadDyno/intercooler-js&lhorie/mithril.js" rel="nofollow">http://www.timqian.com/star-history/#LeadDyno/intercooler-js...</a>