Back and forth we go?
Every time I switch from jruby-1_7 branch to master or vice-versa, I have to do a clean build. Unfortunately a clean build replaces the installed jruby launcher (native bin/jruby) with the bash script, while the bin stubs for gems still reference it via a long shebang. This causes those scripts not to work until I re-install jruby-launcher again.
well, there are two points
I guess the first point is the better solution
Yeah, I understand this is one of the weirder aspects of JRuby. I propose the following:
Installing jruby-launcher alone from here is easy enough, and I believe rvm already does it.
This makes it a one-shot deal to bootstrap into a good dev/test environment.
I'm not sure whether this should be done during the build (copy/link bin/jruby.bash to bin/jruby) or whether we should just always have a bin/jruby present. The issue with the latter is that github will see changes to the file, but .gitignore would avoid that.
I did fix the jruby-dist by telling the archiver to copy bin/jruby.bash to bin/jruby - so that part is now independent of what we do with $JRUBY_HOME/bin/jruby
the jruby-launcher I moved into the bootstrap profile - let's see who complains about that
the creation of bin/jruby is only happening when the file is missing and I could not find any clean tasks which would remove it.
but I do like the github + .gitignore idea since it would relief the build a bit which is unix only tasks anyways.
finally fixes this as well.