Nice. Is this going released as 1.4.2? Also, a blog post denoting the history of why this situation is all messed up would be really nice to distribute. Watching the tweets for "rubygems" (an uptime notifier for me along with general PR watch) about this 1.9 stuff makes me a sad panda.
-Nick On Mon, Jan 3, 2011 at 6:41 PM, Eric Hodel <drbr...@segment7.net> wrote: > I've provided a workaround for RubyGems on 1.9 and committed it. > > It cost us the following features: > > * Legacy remote repository support (fetching /yaml.Z and friends) > > * Setting GEM_PATH and GEM_HOME via ~/.gemrc > > The first feature didn't strictly need to be removed. I just got annoyed > that Gem::SourceIndex loaded Gem::UserInteraction. > > I don't see a way around the restoring the second feature without more hacks. > > In gem_prelude-land rubygems/custom_require is already kinda-sorta loaded. > When loading rubygems.rb we depend on rescuing several LoadErrors, but > gem_prelude is interfering when rubygems.rb is partially loaded. > > The interference causes ruby to try to try load files like > rubygems/defaults/operating_system.rb in rubygems.rb and Win32API in > rubygems/config_file.rb when RubyGems isn't ready to load them from a gem. > > RubyGems::ConfigFile is required to figure out what gem path and gem home to > use. This involves loading YAML as well. > > On the upside, RubyGems should start a little faster now that it no longer > loads rubygems/specification, rubygems/config_file or yaml. > _______________________________________________ > Rubygems-developers mailing list > http://rubyforge.org/projects/rubygems > Rubygems-developers@rubyforge.org > http://rubyforge.org/mailman/listinfo/rubygems-developers > _______________________________________________ Rubygems-developers mailing list http://rubyforge.org/projects/rubygems Rubygems-developers@rubyforge.org http://rubyforge.org/mailman/listinfo/rubygems-developers
