Keep the report lined up even if the label is long
add another time format to Time#parse benchmark
Audited signatures of FFI methods
Bump dev version number
taint not taint!
Adding spec to confirm that Hash#reject taints.
Make the gcc gods happy
Add ability to disable inlining a method preemptively
This fixes Proc.new not seeing a BlockEnvironment because it was running
inside an inlined method with the block inlined (thus no
BlockEnvironment was ever created)
Update website for 1.2.3
Automatically calculate the previous release date
Bump version number to 1.2.3
style reformatting in URI#eql?
made the is_a? check more succinct
using is_a? instead of directly comparing .class
fixed URI#eql? to return false when comparing to non-URI objects
fixed URI#normalize! not downcasing scheme
bench Integer #even/#odd, #times and #upto/#downto
bench added #clone and #new to String dup
bench Array#sort methods
bench String concat methods
bench fix for broken Array#reject!
Be sure to sign extend the result of an FFI call
Merge branch 'doc'
Use Type.obj_kind_of? to a mocked #kind_of? Fixes #762
FreeBSD doesn't need -ldl and OpenBSD already explicitly has it
Store the make to use in a variable and default to gmake on *BSD plat…
Use ugly kludge workaround for weird daedalus bug
Remove bad $~ assignment code
fixes the fixnum benchmarks
Add a few File benchmarks
Add benchmark for String#reverse / String#reverse!
Make sure the string has the same length
The benchmarks tries to expose different comparison times based
on starting / ending with a different character so it shouldn't
trigger the shortcut we have that checks length first.
docs added how-to on benchmarks
bench String#upcase, #downcase and #swapcase