| Age | Commit message (Collapse) | Author | |
|---|---|---|---|
| 2014-04-06 | Remove now unnecessary realpath calls | Jack Nagel | |
| 2014-04-06 | Always resolve one symlink when loading formula files | Jack Nagel | |
| This means that tapped formula will always have the correct path, and we can stop calling realpath everywhere. | |||
| 2014-04-06 | Fall back to open/write when atomic_write fails | Jack Nagel | |
| Fixes #28190. | |||
| 2014-04-05 | Remove obsolete guards around canonical_name return value | Jack Nagel | |
| 2014-04-05 | Implement canonical_name in Formulary | Jack Nagel | |
| 2014-04-05 | Drop support for passing incomplete paths | Jack Nagel | |
| `brew install ./foo` treats `./foo` like `./foo.rb`. This requires a confusing special case in the name resolution logic and doesn't make a whole lot of sense from a UX perspective. Drop support for this. The argument should be an absolute path, a relative path, or a formula name, rather than a hybrid. cf. https://github.com/Homebrew/homebrew/issues/23430#issuecomment-26776962 | |||
| 2014-04-05 | Extract alias resolution to a separate loader class | Jack Nagel | |
| 2014-04-05 | Use StandardLoader when we know the path already | Jack Nagel | |
| 2014-04-05 | Drop readable? check since we don't do that for any other case | Jack Nagel | |
| 2014-04-05 | Simplify tap formula loading | Jack Nagel | |
| 2014-04-05 | Pass path directly to StandardLoader when possible | Jack Nagel | |
| Now we can avoid computing the path twice in the common case. | |||
| 2014-04-05 | Eliminate repeated work in Formulary.factory | Jack Nagel | |
| Much of the name resolution done in Formula.canonical_name is repeated Formulary.factory. Here we eliminate the repeated work by duplicating the code from canonical_name. Later we will refactor it so that both methods can share the bulk of the logic. | |||
| 2014-04-05 | Guard against nil in inherited_options hash | Jack Nagel | |
| Fixes #28188. | |||
| 2014-04-05 | Combine conditional branches that do the same thing | Jack Nagel | |
| 2014-04-05 | FromPathLoader already calls expand_path | Jack Nagel | |
| 2014-04-05 | Prefer File.extname to regexp | Jack Nagel | |
| 2014-04-05 | More correct "symlink already exists" check | Jack Nagel | |
| 2014-04-05 | Avoid realpath where it is not necessary | Jack Nagel | |
| Symlinks in opt and LinkedKegs point directly at a keg in the cellar, so only resolving one symlink should suffice, and make it clear what path we are actually interested in. | |||
| 2014-04-05 | Ignore interrupts while cleaning up | Jack Nagel | |
| 2014-04-05 | Automatically reinstall up-to-date dependencies that are missing options | Jack Nagel | |
| Closes #21223. | |||
| 2014-04-05 | Rename outdated_keg to linked_keg | Jack Nagel | |
| 2014-04-05 | Load install receipt before performing any operations on existing kegs | Jack Nagel | |
| 2014-04-05 | Ensure build deps are installed for deps with inherited options | Jack Nagel | |
| Fixes #28144. | |||
| 2014-04-05 | Add Xcode 5.1.1 (=Golden master build) clang version | Clemens Gruber | |
| Closes #28173. Signed-off-by: Jack Nagel <jacknagel@gmail.com> | |||
| 2014-04-04 | Clean up some concatentation inside interpolation | Jack Nagel | |
| 2014-04-04 | Xcode.prefix: remove osx-gcc-installer hack | Misty De Meo | |
| This hack is starting to outlive its usefulness, and produces false positives. Fixes #28147. | |||
| 2014-04-04 | Need to still check formula build state for some reason | Jack Nagel | |
| 2014-04-04 | Fix universal option inheritance during upgrades | Jack Nagel | |
| 2014-04-04 | python_dependency: don't block pour when :build. | Mike McQuaid | |
| 2014-04-04 | 0644 is not executable | Jack Nagel | |
| 2014-04-03 | Remove need for class name cache by only computing it once per formula | Jack Nagel | |
| 2014-04-03 | python_dependency: only pour with system python. | Mike McQuaid | |
| This is as Python bindings are not build in a way that can have system or Homebrew Python used interchangeably. | |||
| 2014-04-03 | formula_installer: check pour from requirements. | Mike McQuaid | |
| 2014-04-03 | requirement: add pour_bottle? method. | Mike McQuaid | |
| Allows disabling bottles from requirements. | |||
| 2014-04-03 | Build the version scanner regexp once | Jack Nagel | |
| 2014-04-03 | Add missing '`' to message of blacklisted rubinius | Clemens Müller | |
| As pointed out by @mikemcquaid in http://git.io/Q5b9wg Closes #28106. Signed-off-by: Jack Nagel <jacknagel@gmail.com> | |||
| 2014-04-03 | Revert "Audit gnome URLs, prefer download.gnome.org" | Jack Nagel | |
| This reverts commit 557ef8df05a8ea28ed903f08b2009ebc0e218f5a. | |||
| 2014-04-03 | Add keg-only reason symbols for Xcode 4.3 and Xcode 5 | Jack Nagel | |
| Closes #28095. | |||
| 2014-04-03 | Remove special behavior of autotools symbol deps | Jack Nagel | |
| Closes #28094. | |||
| 2014-04-02 | Cache MacOS.version comparison results | Jack Nagel | |
| MacOS.version#<=> is called many, many times during formula loading with the same half dozen or so arguments. A typical call to this method involves: * a hash lookup to convert a symbol argument to a string * creation of a throw-away Version object wrapping the argument * the actual version comparison, which is not cheap This makes it a prime candidate to be memoized. | |||
| 2014-04-02 | Add fast path for CVS dep | Jack Nagel | |
| The search mechanism in MacOS::Xcode is very slow. It requires shelling out at least twice, and possibly a third time (in the CLT-only case). Calling provides_cvs? activates this in order to determine the Xcode version. But if we know that there isn't an Xcode available for the current OS that meets the criteria, we can avoid this check entirely. | |||
| 2014-04-01 | Remove dead code | Jack Nagel | |
| 2014-04-01 | Make keg_only validation lazy | Jack Nagel | |
| 2014-04-01 | The argument to pkgutil_info is always a string | Jack Nagel | |
| 2014-04-01 | Batch up mdfind calls when possible | Jack Nagel | |
| 2014-04-01 | Avoid calling xcodebuild twice | Jack Nagel | |
| 2014-04-01 | Remove unused parameter | Jack Nagel | |
| 2014-04-01 | Extract tag and checksum selection from DSL method | Jack Nagel | |
| 2014-04-01 | Remove now unnecessary branching from bottle DSL methods | Jack Nagel | |
| 2014-04-01 | Stop jumping through hoops to get at the collector object | Jack Nagel | |
