Get Gentoo!
gentoo.org sites
gentoo.org Wiki Bugs Forums Packages
Planet Archives Sources
Infra Status
summaryrefslogtreecommitdiff
path: root/eclass/ruby-fakegem.eclass
Commit message (Collapse)AuthorAgeFilesLines
* Only use revision, not full Id for the eclass.Diego Elio Pettenò2010-02-191-3/+3
|
* Don't create generic wrappers when installing for a single implementation.Diego Elio Pettenò2010-02-191-3/+26
| | | | | | | | | | | | | | | | | | Since we have a few packages that can only be installed for JRuby, and that thus wouldn't work with any of the selected implementations, and a few that installs only for Ruby 1.8 (and thus would break if 1.9 was selected), try to reduce their impact by only producing generic wrappers when installing for multiple implementations. This should produce a totally working system after updating from 1.9, among other things. To make sure that the wrappers generated by different revisions of ruby-fakegem.eclass are properly identified, also add the CVS ID of the eclass in the comments of the generated wrapper. At the same time add the same specification about generation to the gemspec files, so that we can identify “older” specifications quickly.
* Add support for suffixes to a gem (like .java) and support forDiego Elio Pettenò2010-02-191-3/+4
| | | | pre-release mangling (_pre becomes .pre).
* Take into account that the description may contain double quotes and quote ↵Hans de Graaff2010-02-131-2/+3
| | | | them to avoid generating invalid ruby code. Fixes #304723.
* Define each_ruby_test in ruby-fakegem only if a test task is defined.Diego Elio Pettenò2010-01-241-13/+10
| | | | | | | With this change you won't get further output during test phase if there is no fakegem test going to be executed. Packages will still be able to inject their each_ruby_test function to run custom test commands.
* Default to installing the binary wrapper to /usr/bin, when callingDiego Elio Pettenò2010-01-211-2/+4
| | | | | | with just a basename for the wrapped tool. Thanks to Klaus Birkelund Jensen for reporting, in bug #301621.
* Always install the EXTRAINSTALL content, this also allows it to be files ↵Diego Elio Pettenò2010-01-181-8/+9
| | | | rather than just directories
* Correct an undeclared IUSE doc and fix documentation for ↵Diego Elio Pettenò2010-01-091-3/+3
| | | | RUBY_FAKEGEM_REQUIRE_PATHS.
* Don't error out if there is neither bin nor lib directories to install.Diego Elio Pettenò2010-01-011-2/+3
| | | | This is the case of the pg gem for instance.
* Add a variable to explicit further required paths, useful for gems like ↵Diego Elio Pettenò2009-12-261-2/+13
| | | | RedCloth.
* Allow for multiple documentation directories.Diego Elio Pettenò2009-12-211-4/+6
|
* Add a RAKE_FAKEGEM_BINWRAP variable.Diego Elio Pettenò2009-12-201-1/+23
| | | | | This way to just add command wrapping we won't be needing to define a function. This also defaults, the same as original rubygems, to all the binaries installed, but it can be tweaked.
* Change the binary wrapper code.Diego Elio Pettenò2009-12-161-2/+3
| | | | | | | | | | Instead of using the Gem code to find the gem itself (which only worked for gems whose library was named after the gem, and failed for spec and other packages), ask Gem for its system path (which will be dynamic depending on the Ruby implementation used) and affix the hardcoded relative path of the wrapper (which we know at merge time). Tested with rake, bluecloth, rspec, Ruby 1.8, 1.9 and JRuby.
* Fix and extend documentation.Diego Elio Pettenò2009-12-151-3/+14
|
* Use the fakegem name and version vars for the source uri.Diego Elio Pettenò2009-12-151-3/+3
|
* Provide a default SRC_URI.Diego Elio Pettenò2009-12-151-1/+3
|
* Add ruby-fakegem.eclassAlex Legler2009-12-141-0/+264