On Leopard you can enable it by defining the environment variable:
HOMEBREW_USE_LLVM
I didn't enable it on Leopard by default because it doesn't get as much testing, and I don't want to rock the boat. On Snow Leopard people are forigiving of issues because it is new and many compiles fail anyway. And if you installed Snow Leopard you *want* the cutting edge.
ClosesHomebrew/homebrew#29
Otherwise you run the risk of not running the exact version / make of the utility you planned.
FixesHomebrew/homebrew#48
Really we need to do this formula too, so I guess a make and cmake function are on the way…
Eg gettext gets added into LDFLAGS, INCLUDE and that. I hope I got everything
that is typical. Prolly not. But we'll find out.
Made readline keg_only because the BSD version is provided by OS X, and I
don't want bug reports that are tricky to solve due to unexpected differences
between the two.
GNU GetText breaks eg. Ruby 1.9 builds, and some other formula I have been building too. But it is required by eg. glib. So to solve this we are going to by default not symlink gettext into the Homebrew prefix.
Formula that depend on GetText will have the gettext paths added to the brewing environment automatically. Neat.
For when we optimize too much and things seem to break. In my experience the
GCC optimizer is pretty safe nowadays, but at least this allows you to test
the hypothesis the optimization is breaking the build.
Signed Off By: Max Howell <max@methylblue.com>
I realised that -msse4.1 and -msse4.2 aren't supported by GCC 4.0, so I made
the brash decision that we require GCC 4.2. It comes with Xcode 3.1 so people
can upgrade if they have to.
Requiring a single compiler is better for us anyway -- less possible errors
and failures.
Formulae can still request gcc-4.0.1, but at least then those formulae still
only use a single compiler and not possibly two.
This regards Issue Homebrew/homebrew#30.
Turns out -march=native isn't supported by Apple's GCC, but while investigating it I found they'd back ported the -march=core2 option, so we win anyway.
Logic reverted to how it was yesterday.
I moved the gcc options stuff back to brewkit.rb as we manipulate the cflags more later and it seemed bad form to split the logic for this area over two files.
Additionally the brew command exits immediately on powerpc now. Brewkit doesn't throw as theoretically it is a useful library file for other projects.
Simplified hardware model testing.
Even smarter compiler option generation using sysctl and new GCC 4.2 features.
Get processor count from sysctl and thus remove our dependency on RubyCocoa.
Specifying -v/--verbose shows the build environment before the build
MACOS_VERSION contains the floating point value of the OS X version
A test for some floating point assumptions I make
Evidence that using perl from the cli for in-replace is stupid :P
Had to use $'' to allow escaping of ' in bash strings.
Wasn't escaping regexp symbols as well, so it was amazing this worked at all!
strip unlinks the file first, breaking hard links, so we detect instances where we are about to strip a file with many linkages and prevent it.
This fixes the libexec non executable bug in the git package.
Took me a long time to figure out what was wrong! :P
