eventmachine oddities

so if you get ‘frozen Threadgroup’ in mac os x, or if you get ‘hangs’ when you run backticks, it is because Ruby stinks with threads and IO.  Run single threaded and it should work–unless you have tons and tons of ports open, in which case somehow lots of weirdness occurs.  Like lots of port unbinds.

how to get ports to work with leopard

as specified in the port description page, you need to create ~/.profile file  

export PATH=/opt/local/bin:/opt/local/sbin:$PATH
export DISPLAY=:0.0

Then after installing macports directly, you can run
. ~/.profile
on the command line and then run sudo port “whatever”
and it will work now.
And also executables installed by mac port will work as expected, only after running that
. ~/.profile
in each terminal.

ruby mingw win32

So…I tried out Ruby with mingw and it is indeed faster! Significantly faster.  It is like ‘Wow ruby actually works!’

This distro has old version of rubygems, which isn’t quite as compatible, but works as a demo:

http://doachristianturndaily.info/ruby_distro

faster mongrel?

I’ve seen reports of ‘fcgi’ running quicker than mongrel behind nginx. The reporters make the presumption that it is because of unix sockets. So then I wonder if changing evented mongrel to use unix sockets would speed it up (easier to do that than changing mongrel itself, I’d imagine). And perhaps re-using connections would help, as well.

ruby stuffs

test_get_sockname.rb:20: syntax error, unexpected kAND, expecting ‘)’     assert(peername and !peername.blank?)
means “replace your and with &&’s!” NoMethodError: undefined method `assert’ for #<#<Class:0x489e78>:0x489d10> means ‘wrap your asserts with parentheses