escape from getSomeData if there will never be data

[ddyer0]

If the remote end closed the inputstream while the local thread is
in getSomeData, the local end hasn't been closed, and the thread
hangs forever.

[ddyer0]

I discovered this problem incidental to my other recent work on sockets.

[codenameone]

What about the case of buffer.length == 0 ?

[ddyer0]

Buffer length=0 is a legal state, if a previous read exactly emptied the buffer and no more
data has arrived yet. This case never came up before because my previous networking
had reliable unix hosts as the server, which never closed connections unexpectedly. The
current cases involve peer-to-peer connections between tablets, which are much more
vulnerable to unexpected disconnection.

[codenameone]

Length 0 is valid but after this change it will no longer sleep in the loop

[ddyer0]

you're right, add buffer.length==0 && to the if clause

[codenameone]

Thanks!

[ddyer0]

Networking on IOS seems to be completely broken. Unless you changed something else,
this is the immediate suspect, though I probably shouldn't be fixed by simply reverting this.

[ddyer0]

I think the proper fix is to make the IOS implementation conform the the same contract
as android, so in IOSImplementation.java

static final byte[]dummyval = new byte[0]; @Override public byte[] readFromSocketStream(Object socket) { byte []val = nativeInstance.readFromSocketStream(((Long)socket).longValue()); if(val==null) { return(dummyval); } return(val); }
The same problem may exist for javascript

[codenameone]

I reverted the PR for now