Don't reuse same-namespace connections

[rauchg]

io('http://localhost/')
io('http://localhost/')

will create two connections moving forward.

[rauchg]

@michael-luo

[michael-luo]

would the changes be in socket.io-client constructor instead of socket.io because I was calling the socket.io-client constructor in middleware.js with forceNew: true?

i'm wondering if this is the right approach, it works with our insights project when i disable forceNew:

https://github.com/michael-luo/socket.io-client/commit/d409db22420bff14ab6fb0feff5dc3103c810ebb

[rauchg]

Correct. I just keep a single milestone instead of a milestone spread out in many different repos

[michael-luo]

PR: socketio/socket.io-client#808

[cha0s]

Can you clarify why this only applies to the root namespace? It seems weird that we still have to forceNew for other namespaces. Why is '/' special?

[rauchg]

@cha0s it shouldn't be. It should be for any existing namespace that has already been initialized, a subsequent io call will yield a new connection

[rauchg]

The problem that I see with the patch now is that it's not clear where a subsequent new namespace would go to: the first socket or the second

[rauchg]

Actually that's a non-issue since namespaces are completely independent.

[rauchg]

@cha0s in socketio/socket.io-client#808 you should see we're not special-casing /

[cha0s]

Thanks, I misunderstood and thought that https://github.com/michael-luo/socket.io-client/commit/d409db22420bff14ab6fb0feff5dc3103c810ebb#diff-6d186b954a58d5bb740f73d84fe39073R57 was the latest

[peteruithoven]

Could someone elaborate on this choice? This is quite a shift. (I'm not saying it's a bad one.) This means that clients will have to keep a record / cache of their clients themselves from 1.4.0 on.
What happens with the forceNew option? Will there be a option to get a existing connection using another option?
What does moving forward mean? Is this already implemented in the git version of socket.io-client? It's not (yet) documented in the socket.io.client readme.