aboutsummaryrefslogtreecommitdiffstats
path: root/Library/Homebrew
AgeCommit message (Collapse)Author
2013-08-15Add Hardware::CPU.arch_(32|64)_bitMisty De Meo
This replaces hardcoding of i386/x86_64 all over the code.
2013-08-15SubversionDownloadStrategy: fix reference to removed methodJack Nagel
2013-08-14add ubuntu package searchAdam Vandenberg
2013-08-14audit: show real name for aliasesAdam Vandenberg
2013-08-15Unlinked brewed python -> use system pythonSamuel John
Up to now a brewed Python, even if not linked, was preferred over and external Python, so that you had to completely `brew rm python` to have a formula link against external (mostly system) Python. From now on it is okay to `brew unlink python`.
2013-08-14brew reinstall: Reuse options from last timeSamuel John
Now, we can finally stop stuggesting to `brew rm <foo>` and then `brew install <foo> <with-your-preferred options> So `brew reinstall` will honor all options that have been recorded into the INSTALL_RECEIPT.json plus if `--build-bottle` was used.
2013-08-14Keg should link symlinks directly.Samuel John
If the `src` in the Cellar of a formula is a symlink to another dir in the same Cellar (for example `Frameworks/QtGui.framework` is a symlink to `lib/QtGui.framework`), then we should make a symlink directly instead of possibly creating an empty dir. This is needed because Ruby's `find` does not follow symlinks (newer versions can, but we have to support 10.8 for a long time). Currently this probably only occurs for Qt, where we put symlinks into the Cellar/Frameworks and the frameworks linking code is a bit special to create a directory to "merge" together different versions of the same framework from different formulae. See `python` and `python3`; both install `Python.framework` and only one can be the `Current` but the other should still be put into `Python.framework/Versions`. Therefore `Python.framework` has to be a dir and no symlink.
2013-08-14Correct commentJack Nagel
2013-08-14Use each instead of while loopJack Nagel
2013-08-14Replace long conditional with guard clausesJack Nagel
2013-08-14Iterate over these directly rather than mapping firstJack Nagel
2013-08-14Return value of modify_build_environment is unimportantJack Nagel
2013-08-14Reverse conditionalJack Nagel
2013-08-14Collapse to one-linerJack Nagel
2013-08-14No need to map, all? takes a blockJack Nagel
2013-08-14Add missing falseJack Nagel
2013-08-14Remove stray spaceJack Nagel
2013-08-13pathname: remove unused return valuesAdam Vandenberg
Return value was used only by one test, rewrote test to know expected value.
2013-08-13Remove outdated commentJack Nagel
2013-08-13Avoid reading whole files into memory during decompressionJack Nagel
2013-08-13Improve checksum perf by providing an output bufferJack Nagel
2013-08-10Doctor: missed CLT URL replacementMisty De Meo
2013-08-10Update URL for CLT downloadMisty De Meo
http://connect.apple.com is still down, but https://developer.apple.com/downloads has all of the Apple developer downloads available. Fixes #21812.
2013-08-10Check that the cellar exists in Formula.installedJack Nagel
Fixes #21741.
2013-08-10Make usage of ObserverPathnameExtension more obviousJack Nagel
Remove use of globals. Closes #21795.
2013-08-10Allow specifying a custom PATH for which methodJack Nagel
Closes #21794.
2013-08-09Always prune empty toplevel directories when unlinkingJack Nagel
Closes #21750.
2013-08-09Keep track of pruned parent directoriesJack Nagel
2013-08-09Add tests to demonstrate failure to prune empty directoriesJack Nagel
2013-08-09Remove hardcoded paths in keg testsJack Nagel
2013-08-09Remove keg test dependency on formulaJack Nagel
2013-08-09Move MacOS.xcode43_without_clt? into Xcode moduleJack Nagel
2013-08-09Add fedora to brew searchAdam Vandenberg
2013-08-08alphabetize listAdam Vandenberg
2013-08-08Add SourceForge hg patternAdam Vandenberg
2013-08-08cleaner: traverse directories in postorderJack Nagel
2013-08-08exception styleAdam Vandenberg
2013-08-08Don't hard-code PYTHONPATH for `brew test`Samuel John
This rules out any possibility to test Python 3.x software as the PYTHONPATH is used by any Python and `brew test pyqt --with-python3` fails because it picks up the `pyqt` module fomr Python 2.x's site-packages. Upstream tells us not to set the PYTHONPATH and then starting another (major) python version.
2013-08-08Improve docs for Formula#python (python_helper)Samuel John
2013-08-08Python: Exit if PYTHONPATH is wrongSamuel John
In our sitecustomize.py we instruct python to exit with an error message if the PYTHONPATH is pointing to a wrong homebrew site-packages dir. With wrong meaning another major python version. If you set the PYTHONPATH to include $(brew --prefix)/lib/python2.7/site-packages and start python3, it may pick up modules from there, wich can result in errors for non pure python modules (such as PyQt).
2013-08-08For brewed python, make sure to unset the PYTHONPATHSamuel John
during building of python software inside of a `python do … end` block.
2013-08-08python_helper prepends PYTHONPATHSamuel John
…instead of appending. This should be safer against any malformed user-defined values in the PYTHONPATH.
2013-08-06Use MacOS::X11 path helpersJack Nagel
2013-08-06Eliminate MacSystem moduleJack Nagel
2013-08-06MacSystem now just wraps MacOSJack Nagel
2013-08-06Remove more dead codeJack Nagel
2013-08-06Don't use exceptions for flow-controlJack Nagel
2013-08-06Raise a useful messageJack Nagel
2013-08-06styleJack Nagel
2013-08-06Make this logic less confusingJack Nagel
generated by cgit v1.2.3 (git 2.25.1) at 2026-03-07 14:39:00 +0000