Closes #149.

Signed-off-by: Andrew Janke <andrew@apjanke.net>

Provide a single command that can be run to disable analytics, run it if
`HOMEBREW_NO_ANALYTICS` is ever set and remove the user UUID file in
that case too.

References https://github.com/Homebrew/brew/issues/142.

Make sure that users are notified on the first run of `brew update`
after we enabled analytics about how it works and how to opt-out. This
will be shown to all users who have not already seen this message from
`brew update` or through a new Homebrew installation.

References https://github.com/Homebrew/install/pull/42
References https://github.com/Homebrew/brew/issues/142

Sometimes they are in e.g. libexec but there's no real reason to assume
they are anywhere; we want to relocate them regardless.

For more information and opt-out instructions please read:
https://github.com/Homebrew/brew/blob/master/share/doc/homebrew/Analytics.md#analytics

Closes #118.

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

This will be expanded to a full expat dependency on Tiger alone.

Fixes mistydemeo/tigerbrew#30.

Setting -mcpu and -mtune on Tiger with gcc-4.2 exposes a bug in one of the
system headers, causing certain builds to fail. This can be fixed by
adding -faltivec to CFLAGS.

See: http://trac.macports.org/ticket/34213

Fixes mistydemeo/tigerbrew#207.
Fixes mistydemeo/tigerbrew#330.

Fixes mistydemeo/tigerbrew#37.

Fixes mistydemeo/tigerbrew#365.

The new stage() signature introduced by #66 breaks back-compatibility
under Ruby 1.8.7. This fixes it by switching back to a one-argument
block signature and using a new class to wrap both the Resource and
Mktemp info for the staging context, in a signature-back-compatible
way.

Addresses homebrew/homebrew-core#529.

Closes #135.

Signed-off-by: Andrew Janke <andrew@apjanke.net>

We want to always avoid merge commits being committed to Homebrew so
this feels like a better fit.

Make multiple replacements if using the block form.

It simplifies the number of commands and it's now an officially
supported command again.

Closes #119.

Signed-off-by: Martin Afanasjew <martin@afanasjew.de>

Since switching from `system` to `Utils.popen_read` we no longer need to
suppress Ruby's output of `Syntax OK` to `$stdout`.

Previously, syntax warnings were printed, but didn't cause `readall` to
exit with a non-zero exit code. Now they do, making it easier to catch
accidentally introduced syntax warnings in the test bot.

The current one returns a 404.

Reorders the PR submission steps so `brew audit` is done after `brew install`,
so it can pick up issues with the installed formula, like non-executable stuff
in `bin/`. Matches the `test-bot` process better.

Minor style changes, including consistently using periods at the end of list items
which are complete imperative sentences.

Closes #127.

Signed-off-by: Andrew Janke <andrew@apjanke.net>

Makes identifying which line in a formula has the bad use of 'option' easier.

Closes #120.

Signed-off-by: Andrew Janke <andrew@apjanke.net>

This collects all violations for each formula in a single place, instead
of doing `brew style` outputs for all formulae first, and then the other
audit checks.

Closes #112.

Signed-off-by: Andrew Janke <andrew@apjanke.net>

Closes #125.

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

This version is never wanted at this point and it will help Homebrew
Cask deal with the annoying errors that result from having this version
still around (some which I've already help users debug).

Someone else has been sending weird data to ours and it has bad initial
test data so let's roll it before we ship it to everyone on Saturday.

Closes #114.

Signed-off-by: Martin Afanasjew <martin@afanasjew.de>

Use the `reason` attribute to be able to handle `UsageError` subclasses
more uniformly and simplify logic in `brew.rb` to handle them together.

Instead of always printing the generic help text, print command-specific
help if it is available and a command raised the `UsageError` exception.
Put the error message underneath the help text (was above) to avoid that
it scrolls off the screen.

Thereby fix a regression where handling the invalid usage would fail to
access `ARGV.usage` removed in c6536066.

Turns out making `empty_argv` a boolean argument for `Homebrew.help` was
not the best idea and having command-to-path mapping and help extraction
in a single method is not flexible enough.

Also only complain about missing help text when `HOMEBREW_DEVELOPER=1`
and otherwise just print the generic help text.

Format the usage examples more consistently (particularly parentheses
instead of square brackets where one of several alternatives has to be
picked). And add the now much more useful `brew help <command>`.

Closes #113.

Signed-off-by: Martin Afanasjew <martin@afanasjew.de>