We don't want bug reports about this crap, even though the user REALLY should listen to the warning message. What actually happens is they blame us instead.

This kind of warning in particular:

warning: Insecure world writable dir /usr/local/bin in PATH

You have to hold people's hands, they won't google anything.

Refs Homebrew/homebrew#13982.

Now that X11 components are specified as dependencies, users will be
prompted to install these components when necessary, and this check is
no longer needed.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Tell user exactly what command to use rather than thinking they'll figure it out by applying some negation logic to the main verb in the sentence they just read.

Closes Homebrew/homebrew#12790.

Fixes Homebrew/homebrew#12446.

About time right?

Removes dead symlinks in Formula and re-symlinks all Taps.

Closes Homebrew/homebrew#11565.

Repos can't at this point. Hard to easily fix so avoiding it until 2.0 or until someone moans enough.

Fix Homebrew/homebrew#11728, fix Homebrew/homebrew#12890, fix Homebrew/homebrew#13653, fix Homebrew/homebrew#13734, fix Homebrew/homebrew#13787, fix Homebrew/homebrew#13818,
fix Homebrew/homebrew#13939.

It's also necessary to delete LC_ALL or it overrides the new LANG.
LC_COLLATE is the problem variable but is affected by LANG and LC_ALL
so just use them instead. Extends the fix made to geoip in dc955c.

Fix Homebrew/homebrew#11728, fix Homebrew/homebrew#12890, fix Homebrew/homebrew#13653, fix Homebrew/homebrew#13734, fix Homebrew/homebrew#13787, fix Homebrew/homebrew#13818.

Possibly would help diagnose: Homebrew/homebrew#13853.

Because, if you do, your changes will be lost.

In order to better support Xcode-only systems, where X11 libs and
executables live under /usr/X11 but headers live in the SDK, move the
x11_* helper methods into a new module.

This allows us to keep some of the CLT/Xcode-only and Apple X11/XQuartz
logic hidden from outside code, like ENV.x11.

Since Apple's X11 is actually XQuartz, name the module "MacOS::XQuartz".

Since we're splitting on newlines, stripping shouldn't be necessary.

Turns out mdfind can return multiple results on the commandline, if
more than one app bundle matches.

Fixes Homebrew/homebrew#13789.

Add brew/science.

Fixes Homebrew/homebrew#12779. Closes Homebrew/homebrew#12784.

utils.rb invokes bash as a non interactive shell to run the editor
when you run brew edit.

Non-interactive shells are not intended to capture user input and for running scripts. Whilst the invocation of the editor is a script the actual editor is intended for accept user input so this is an issue with brew and not emacs.

This causes issues with emacs which has different behaviors when not running under an interactive shell.

Signed-off-by: Max Howell <mxcl@me.com>

conflicts_with is a thin wrapper around Requirement which simplifies
marking conflicts between formulae.

Closes Homebrew/homebrew#13687.

Signed-off-by: Misty De Meo <mistydemeo@gmail.com>

10.8 no longer ships four .pc files which were included in previous
versions, though it does still include the libraries which they pointed
to. This commit provides copies of the .pc files for all of these, and
prepends Library/Homebrew/pkgconfig to PKG_CONFIG_PATH on 10.8.

The .pc files in question are: lib curl, lubexslt, libxml-2.0, libxslt

Fixes Homebrew/homebrew#13673.
Fixes Homebrew/homebrew#13627.
Fixes Homebrew/homebrew#13652.
Fixes Homebrew/homebrew#13482.
Closes Homebrew/homebrew#13572.

Signed-off-by: Misty De Meo <mistydemeo@gmail.com>

Keg#link would sometimes count a linked file when doing mkpath, even if
the target directory already exists; #unlink would never count it. This
meant that "brew ln" and "brew unlink" counts for the same keg could be
out of sync with each other.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Depending on `open-mpi` or `mpich2` will cause problems since both formulae
install components with the same names. `brew audit` now recommends using
MPIDependency.

Signed-off-by: Charlie Sharpsteen <source@sharpsteen.net>

All versions of OS X prior to Lion shipped with some version of Open-MPI, but
without working compiler wrappers for Fortran. Homebrew currently has two
formulae that can supply this software: open-mpi and mpich2.

This commit adds a `MPIDependency` Requirement that can be passed one of four
values when constructed:

    :cc, :cxx, :f90, :f77

This will ensure the `mpi<value>` compiler is available and working. For
example, if `depends_on MPIDependency.new(:cc, :f90)` is used, the Requirement
will search for working `mpicc` and `mpif90` wrappers.

If the required wrappers cannot be found, an error message will be displayed
that prompts the user to install one of the Homebrew formulae that provides
MPI.

For each language passed to a MPIDependency Requirement, environment variables
be will set that point to the compilers.

Signed-off-by: Charlie Sharpsteen <source@sharpsteen.net>

Because of some quirks in how formulae are loaded, DSL methods invoked
in the class definition are called multiple times. This results in, for
example, duplicate dependencies and requirements.

Code that iterates over requirements and takes some action for each can
thus repeat things that shouldn't be repeated, like appending to
environment variables.

Make DependencyCollector's external_deps a Set, and define eql? and hash
on Requirement to prevent these duplicates.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Rather than doing type introspection in build.rb, just define a method
to perform the necessary environment setup for Requirements.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

MacOS.sdk_path is meant to return the SDK path that matches the version
argument, so store the result in a hash.

Fixes Homebrew/homebrew#13623.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Following feedback from @mxcl on bcda0d.

The fish completion hadn't been updated since 2009; this brings it up
to date. Changes include:

* Create the list of commands the same way as in bash completion
* Add every option for every current command, with descriptions
* Fixes a bug in fish_complete_brew_command
* Allow fish_complete_brew_command to take multiple commands
* Expand functionality of fish_complete_homebrew_formula to support
  other commands and taps, and change name to reflect new functionality

Tested on fishfish, but should be compatible with fish.

Fixes the problems discussed in 34c970 to ensure that `brew doctor`
does not look like it is broken on user setup problems but still
returns a sensible exit code for e.g. BrewBot.

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Signed-off-by: Jack Nagel <jacknagel@gmail.com>

Signed-off-by: Jack Nagel <jacknagel@gmail.com>