Activity (RSS)

I've refactored the code since I posted it last night.

I've removed the need for --cluster opting for --no-config on the cluster spawns since all of its optio...

hey thx for the patch Matt! I got a couple questionsabout it.

I'm not sure using the config file should be the default. Mostly the config file will be used on...

Might want to change the --config option to accept a variable param and be defined as --[no-]config to get rid of the "no" value to use no config file.

Here is the patch file.

To create a configuration file:

thin configure --other-opts...

The --config/-C option used here tells thin where to put the configu...

could we catch the ThreadError and see which thread threw the error? If not the thread Thin cares about, ignore it?

just thinking out loud

Porting to C wasn't worth it. I altered Thin::Headers to output directly to a string rather then store into an Array that was latter rewritten into a string.

Also improved specs stability. Should not leave phantom processes after running specs now.

How many ppl are gonna use this?

I'm reluctant to do to this because it will require maintaining twice more code, in 2 languages. Mongrel already has a Java p...

discussed/requested on #thin

need ragel to output java, need to port thin.c to java(?), need to use either pure ruby version of EM or new java version of EM, ...

Now moves request body to tempfile if > 112 MB (like mongrel)

If request body is less then 10kb put in a stringio if its over 10kb put it in a tmpfile

After some profiling Thin::Headers is taking lots of time.

Maybe port this to C or optimize the code. rassoc might not the best choice there.

Port response writing to a C extension

can't reproduce, I'll reopen if needed

Carl Mercier is experiencing weird stuff, response headers seems to be happened to the top of every HTML page in the browser ?wtf?

not a ticket, but would be cool if someone ran some benchmarks for those

Add a better god sample file in example/, maybe generate one w/ thin config and use this in replacement of mongrel_cluster to launch cluster at startup and all...

should use the same format as mongrel_cluster if possible to ease transition.

I don't plan to work on this now, that would be a very much appreciated contribu...

Unless someone already has some code for this or some specific requirements, I should be able to get something put together this weekend.

I assume YAML would ...

Yeah boolean option.

It would write HTTP requests and responses to the logfile when turned on + some backtrace. Would be less verbose then --debug.

Does this option set a file to log to? What exactly is returned? When I think "trace", I don't exactly think of logging (though that's either my own semantic d...

request/response logging for debugging purposes

Added to EventMachine trunk, assume it will be in the next version.

http://rubyforge.org/pipermail/e...

I don't think we have anything to do for this, else then wait for EM next release.

Looks like it's on trunk:

http://groups.google.com/group/e...

(should be able to bind to ::1)

Change benchmarking tests to use:

http://blog.evanweaver.com/files...

Would love to be able to use dtrace / Instruments to profile it

Check for memory usage/leaks using dike or other:

http://blog.evanweaver.com/artic...

Thin uses more memory then Mongrel on some apps (refactormycode), investigate why

Create a Stats adapter to show server stats in the browser

Calling unlock twice on a mutex raises an ThreadError which crashes Thin.

See http://www.ruby-forum.com/topic/...

Not sure we can do anything about this.