Fixes the node start IPC connection

[frozeman]

This PR is adding a loop to test an IPC connection. When a connection timeout is set, it tries to connect every 200ms to the connection, if not it fails.

This speeds up the connection process, as we don't have to wait arbitrary 4 seconds anymore (which weren't enough for eth)

@hiddentao This PR works, but if you set a wrong IPC path it doesn't quit gracefully, instead i delivers an uncaught exception. Could you please look into this.
Simply let the eth node start (Change the node files content in ~/Library/App..Support/Mist/node to eth)
Before change the pic path in the getIPCPath.js to some other path.

You will see it fails, but not gracefully, as the error seems to travers through and doesn't get catched.

This error is created in the socket.js line 84

[2016-06-16 16:44:28.827] [INFO] Sockets/node-ipc - Connect to {"path":"/Users/frozeman/Library/Ethereum/gethdddd.ipc"}
[2016-06-16 16:44:38.965] [ERROR] Sockets/node-ipc - { Error: connect ENOENT /Users/frozeman/Library/Ethereum/gethdddd.ipc
    at Object.exports._errnoException (util.js:949:11)
    at exports._exceptionWithHostPort (util.js:972:20)
    at PipeConnectWrap.afterConnect [as oncomplete] (net.js:1080:14)
  code: 'ENOENT',
  errno: 'ENOENT',
  syscall: 'connect',
  address: '/Users/frozeman/Library/Ethereum/gethdddd.ipc' }
Error: connect ENOENT /Users/frozeman/Library/Ethereum/gethdddd.ipc
    at Object.exports._errnoException (util.js:949:11)
    at exports._exceptionWithHostPort (util.js:972:20)
    at PipeConnectWrap.afterConnect [as oncomplete] (net.js:1080:14)
[2016-06-16 16:44:38.967] [ERROR] main - UNCAUGHT EXCEPTION { Error: connect ENOENT /Users/frozeman/Library/Ethereum/gethdddd.ipc
    at Object.exports._errnoException (util.js:949:11)
    at exports._exceptionWithHostPort (util.js:972:20)
    at PipeConnectWrap.afterConnect [as oncomplete] (net.js:1080:14)
  code: 'ENOENT',
  errno: 'ENOENT',
  syscall: 'connect',
  address: '/Users/frozeman/Library/Ethereum/gethdddd.ipc' }
Error: connect ENOENT /Users/frozeman/Library/Ethereum/gethdddd.ipc
    at Object.exports._errnoException (util.js:949:11)
    at exports._exceptionWithHostPort (util.js:972:20)
    at PipeConnectWrap.afterConnect [as oncomplete] (net.js:1080:14)

[frozeman]

We need to increase this to 30s again after its fixed

[hiddentao]

I will refactor the interval timer loop.

[hiddentao]

Changes:

• Commented out the master password window and logic, eth now just works
• Refactored the socket connection code, so that it waits for a result before re-connecting
• Now fails gracefully
• Added node and network cli parameters

[hiddentao]

@frozeman let me know what you think.

[hiddentao]

Also noticed that the splash screen is showing state download info, but for Eth there isn't any. This needs to be removed before this PR is complete.

Update: now fixed.

[hiddentao]

Only pending issue - should we just remove the master password logic from the node startup?

[hiddentao]

Final changes made, ready for review.

[alexvandesande]

What's the status of this? Is it going in 0.7.6?

[hiddentao]

This is done and ready for review. It should definitely go into 0.7.6.

[hiddentao]

To test eth startup:

$ electron . --loglevel debug --node eth --network test

[frozeman]

we should remove the master password logic. And provide a script for people to get rid of it

[hiddentao]

It's removed.

[luclu]

Seems to work well - both windows 7 and OSX.
Tested starting, switching networks/nodes, detaching.
But I had to remove eth's config folder.


[lock]

This thread has been automatically locked because it has not had recent activity. Please open a new issue for related bugs and link to relevant comments in this thread.