Activity (RSS)

Doing

sudo gem install eventmachine --source http://code.macournoyer.com

should install EventMachine 0.11.0

check you have 0.11.0:

gem list -l eventmachin...

I installed eventmachine 0.10 but still the same issue.

Is it because I'm chrooted?

what is supposed to give epoll? glib?

that is weird, looks like epoll is not available or something.

Can you try to install latest version of eventmachine?

sudo gem install eventmachine --source ...

I'm trying to use Thin as my rails webserver on a Synology NAS.

the server is put on a chrooted debian.

When the server is started, I have the following erro...

implemented in http://github.com/macournoyer/th...

we talked about something like this on IRC:

app.deferred?

If app respond to deferred? and deferred(env) returns true the request will spawn a thread an call...

Allow setting which URL require a thread. Those requests would be processed inside a thread, but not the other requests.

Would be cool to have a common API w/...

--backend option added in http://github.com/macournoyer/th...

thx for the suggestion Brian!

In threaded mode add an option --max-threads to limit the number of active threads. When that number is reached serve sequentially

now auto detects for rails, ramaze, merb and halycon

can also specified w/ --adapter option

fixed in http://github.com/macournoyer/th...

Mongrel does have page caching activated all the time, but it's only for GET and HEAD request, which is the issue in this case.

I'll fix the Rails adapter in ...

Thin is hard coded to require page caching. Mongrel doesn't do this, so thin is incompatible with some configurations of existing mongrel installations.

I'd l...

Never mind. It looks like its already done.

I have a couple of apps that use comet or long polling.

To do this, I needed to subclass Connection. Unfortunately, Connection#process has a long implementati...

Added threaded option in http://github.com/macournoyer/th...

Hey Ry,

Keep alive is the default behaviour in HTTP 1.1 unless Connection: close is specified. Your application must set Content-Length 'cause some browsers w...

hi marc,

I can't get keep-alive to work in either 0.7 or HEAD. I'm doing something simple like this:

app = SimpleApp.new

server = Thin::Server.new('0.0.0.0'...

Sounds good to me.

proctitle for thin has been implemented by grempe in a rack middleware: http://github.com/grempe/thin-pr...

what about a --backend NAME option

that would call: Thin::Backends::.new(host, port)

let me know what you think

We would like to author a custom backend for Thin and be able to register a configuration option indicating that it should be used.

Since http://github.com/macournoyer/th... error is just outputted to console without backtrace of uncaught exception.

Reopen if you still see an issue

I just implemented something basic in http://github.com/macournoyer/th...

See lib/rack/adapter/loader.rb for details. Send me a patch if you can add support f...

I'll definitely contribute to this when the details get ironed out a bit, not only for my own framework but for others as well.

Hey Aníba,

What behaviour did you expect when running config without -C ?

li19-22:/etc/thin# thin -v

thin 0.7.1 codename Fancy Pants

li19-22:/etc/thin# thin config -S /home/mongrel/tmp/thin.sock -e production -c =/var/www/to2blogs/...

I got specs consistently passing on both Mac OS and Ubuntu. I'll try to setup other VMs and check on those

Another option to consider: framework-specific binaries.

thin_rails

thin_merb

thin_ramaze

etc.

I'm not saying it's a better idea, just another idea.

Awesome, the spec passes for me too. Thanks!

I pull you commit in http://github.com/macournoyer/th...

And added a spec example, let me know what you think

I forgot to add, yes I did try to add a spec for this, but my brain isn't working right with all the forking and I couldn't get the spec to work properly (I'm ...

As discussed in http://groups.google.com/group/t..., Thin should check for stale PID files and delete them instead of simply failing.

I've sent a Github pull ...

oops forgot to close

The last Daemonizing failyre is due to the previous crash I presume, cause the pid file is still there.

When you say Ruby crash, you mean the: std::runtime_er...

I'm guessing it's because we start/stop EM multiple times in the specs. When I run each spec files individually I never get those.

The EM trunk has the "std::runtime_error" errors fixed, or so they say. That's why I'm running EM trunk.

I just sync'd and re-ran 'rake spec'. I get one error on the "Thin::Server on TCP socket should GET from Net::HTTP" spec saying "Connection reset by peer". I'm...

The thin w/ those crashed is that it's impossible to rescue those errors. I submitted a ticket to EM: http://rubyeventmachine.com/tick...

that's crazy, I get no failures and no zombies on my Mac OS 10.5. But I do get those "std::runtime_error" sometimes which are due to EventMachine crashing, I'v...