Ramp server reuses promises

[cjohansen]

It seems the capture server will try to resolve already resolved promises under some conditions.

To reproduce

• Fire up the server
• Capture IE8
• Run a test file that contains var document = {};
• The test run should halt (this is probably another bug...)
• Ctrl-c the run
• Go to IE8, hit refresh
• Watch the server crash

Here's the trace I got:

$ buster server
buster-server running on http://localhost:1111

/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/when/when.js:246
                throw new Error("already completed");
          ^
Error: already completed
    at alreadyCompleted (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/when/when.js:246:11)
    at Object.resolve (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/when/when.js:189:4)
    at Object._onImprisoned (/home/christian/projects/busterjs/modules/buster-capture-server/lib/slave.js:114:36)
    at /home/christian/projects/busterjs/modules/buster-capture-server/lib/pubsub-client.js:82:47
    at Object.trigger (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:383:19)
    at Object.distributeMessage (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:666:30)
    at Object._deliverMessage (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:1068:20)
    at Object.<anonymous> (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:1007:12)
    at Object.pipeThroughExtensions (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:529:44)
    at Object.receiveMessage (/home/christian/projects/busterjs/modules/buster-capture-server/node_modules/faye/node/faye-node.js:1003:10)

[augustl]

Thanks, only had a stack trace from someone else, didn't know how to reproduce.

[samgiles]

I seem to encounter your first issue 'The test run should halt (this is probably another bug...)' quite consistently when running the unit tests from https://github.com/ftlabs/ftcolumnflow with buster over a WiFi connection with an iPad 5.0.1 device with Safari 5.1. The test it halts on seems to vary.

What I did notice with the above is that the slave refreshes during the tests, and then the tests halt. But like I said it's strange because the test that it halts on seems to vary, for example sometimes it passes (so I don't think it's the tests).

I wish I could take a look to help further as I've began to encounter both these bugs. If you need anymore information let me know, I'm going to see if I can see anything in the meantime.

Cheers

Sam

[augustl]

Fixed in latest HEAD.