1.4.0 is backwards incompatible and causing karma to fail

[TheSavior]

1.4.0 caused a change that is making karma fail.

See karma-runner/karma#1782 for more context

[deltreey]

👍 I'm seeing this also

[ryanmurakami]

I can also confirm that v1.3.7 works fine with karma.

[ekryski]

Yes I am seeing this in Feathers. It looks like io.sockets.sockets is returning an object now.

example:

var socketio = require('socket.io');
var io = socketio.listen();

// Now returns an object whereas v1.3.7 returned an array
console.log(io.sockets.sockets);

If this is a necessary breaking change then it should probably be a major release? 😄

[ekryski]

Looks like it is this commit b73d9be. I've fixed it in our lib by just wrapping our call to get the socket clients in Object.keys(io.sockets.sockets).forEach(...).

[deltreey]

what a strange commit...no reason listed just "here, arrays are now objects" Wonder what decision led to that.

[bencroskery]

An odd change, but it does seem to make the code a little easier to read (in my opinion) which I guess is what led to the change.

I ended up fixing my issue by wrapping the call like this _.values(io.sockets.sockets).forEach(...) using lodash so I could access the individual socket properties.

[ekryski]

@rauchg responded to me on twitter that it had huge performance improvements. I'm not surprised there.

[edmorley]

Please can you unpublish v1.4.0 and re-release as v2.0.0? Like others above this caused breaking changes with karma, and whilst we can work around it, it would avoid others having to do the same. Thanks :-)

[tangentfairy-zz]

I'm also having an issue with Karma failing since this morning, same forEach error 👍

[kayz1]

Same here. Replaced with _.values(io.sockets.sockets)

[dspnorman]

Please unpublish v1.4.0 and re-release as v2.0.0. Breaking changes in a minor version are really disconcerting.

[darrachequesne]

Hi! I totally understand the concerns here, and I let @rauchg be the judge on whether unpublishing 1.4.0, but... isn't io.sockets.sockets object supposed to be socket.io's internals ? Shouldn't one use io.sockets.clients method instead ?

[ekryski]

@darrachequesne We are actually using that method. I just used io.sockets.sockets to show where it was originally coming from.

[milesingrams]

I encountered a similar Array > Object problem with socket.rooms that stopped my application from functioning correctly. It should definitely be marked a major release (v2.0.0).

[mikermcneil]

In case it helps, here is a quick rundown of what we had to address before updating Sails yesterday: balderdashy/sails-hook-sockets@95b2506

To be fair, we were using undocumented features.

[mikermcneil]

isn't io.sockets.sockets object supposed to be socket.io's internals ? Shouldn't one use io.sockets.clients method instead ?

👍

[davidreher]

None-the-less breaking changes should be released in a new major version. http://semver.org/

[full-of-foo]

I'm also encountering this too.

Chrome 47.0.2526 (Mac OS X 10.11.2): Executed 135 of 135 SUCCESS (1.61 secs / 1.507 secs)
Missing error handler on `socket`.
TypeError: sockets.forEach is not a function
    at disconnectBrowsers (/Users/lol/test/node_modules/karma/lib/server.js:311:13)
    at null.<anonymous> (/Users/lol/test/node_modules/karma/lib/server.js:291:7)
    at emitTwo (events.js:92:20)
    at emit (events.js:172:7)
    at emitRunCompleteIfAllBrowsersDone (/Users/lol/test/node_modules/karma/lib/server.js:256:12)
    at null.<anonymous> (/Users/lol/test/node_modules/karma/lib/server.js:278:9)
    at emitTwo (events.js:92:20)
    at emit (events.js:172:7)
    at onComplete (/Users/lol/test/node_modules/karma/lib/browser.js:142:13)
    at Socket.<anonymous> (/Users/lol/test/node_modules/karma/lib/events.js:13:22)
    at emitTwo (events.js:92:20)
    at Socket.emit (events.js:172:7)
    at Socket.onevent (/Users/lol/test/node_modules/karma/node_modules/socket.io/lib/socket.js:335:8)
    at Socket.onpacket (/Users/lol/test/node_modules/karma/node_modules/socket.io/lib/socket.js:295:12)
    at Client.ondecoded (/Users/lol/test/node_modules/karma/node_modules/socket.io/lib/client.js:193:14)
    at Decoder.Emitter.emit (/Users/lol/test/node_modules/karma/node_modules/socket.io/node_modules/socket.io-parser/node_modules/component-emitter/index.js:134:20)
Chrome 47.0.2526 (Mac OS X 10.11.2): Executed 135 of 135 DISCONNECTED (11.588 secs / 1.507 secs)
/Users/lol/test/node_modules/karma/lib/server.js:311
    sockets.forEach(function (socket) {
            ^

TypeError: sockets.forEach is not a function
    at disconnectBrowsers (/Users/lol/test/node_modules/karma/lib/server.js:311:13)
    at process.<anonymous> (/Users/lol/test/node_modules/karma/lib/server.js:352:5)
    at emitOne (events.js:77:13)
    at process.emit (events.js:169:7)
    at process._fatalException (node.js:211:26)

[levino]

I think this also might break socket.io-adapter@0.4.0 Am I wrong?

[dignifiedquire]

This is fixed in the latest version of karma. It was due to karma using socket.io internals.

[oniono]

NAVER - http://www.naver.com/

oniono@naver.com 님께 보내신 메일 <Re: [socket.io] 1.4.0 is backwards incompatible and causing karma to fail (#2368)> 이 다음과 같은 이유로 전송 실패했습니다.

---

받는 사람이 회원님의 메일을 수신차단 하였습니다.

---

[woozyking]

I encountered the same issue as @milesingrams when checking whether given socket has joined a room yet. The change is good in a sense that it's much faster to do the check than an array indexOf. But I completely agree with everyone above that says about following semver.

However, seeing that there are 5 z versions released already after 1.4.0, republish as 2.0 might not be as simple as 2 months ago, I would suggest to just clearly document it in at least README, preferably in the first few lines to warn all users.