webrick hangs sometimes with guard-jasmine cli

[Huy Nguyen]

I'm having trouble getting guard-jasmine cli working consistently on a jenkins ci server.

When running

`bundle exec guard-jasmine`

```bash

Guard::Jasmine starts Rack test server on port 8888 in development environment.

[2011-12-18 19:14:16] INFO  WEBrick 1.3.1

[2011-12-18 19:14:16] INFO  ruby 1.9.3 (2011-10-30) [x86_64-linux]

[2011-12-18 19:14:16] WARN  TCPServer Error: Address already in use - bind(2)

[2011-12-18 19:14:16] INFO  WEBrick::HTTPServer#start: pid=14343 port=8888

```

It will just hang and never run the spes, the odd thing about it is I can issue requests to the server, via `curl http://localhost:8888` so it's running. Killing it and running it again sometimes the specs will run as expected. Also `guard` by itself will always works and I can't reproduce this on my local env. I tried to trace this a little with put statements in server.rb and it looks like once the Rack/Webrick thread gets to that point it never returns execution to the main thread?

Could this be related? http://bugs.ruby-lang.org/issues/4841

ci server is running ubuntu 11.04

Any ideas would be appreciated.

[Michael Kessler]

I'm also having some troubles with the server start in guard-jasmine, it seems a bit unstable, but I can't figure out the problem. I just released guard-jasmine, which allows you to use also mongrel or thin as testing server:

guard :jasmine, :server => :thin do
...
end

Now you can test if your issue is related to WEBrick or not.

> <Gemfile.txt>

[Huy Nguyen]

Swapped it out with thin and got the same result. I'm not sure what it might be, I tried to do a manual sleep instead of checking for server up but as soon as the server thread fires it never comes back. If I have time I'm going to try and rewrite parts of server.rb to match capybara's server.rb to see if makes any difference.