Hi,
end of last week I finally found the time to finish the port of
gem_install.sh to a ruby script. It is still called .sh but it
is only a shim shell wrapper which finds all installed ruby
versions and then calls the script with each ruby version.
Why all the work? In many cases our old gem_install.sh could only
do its work based on assumptions. [1] It had no way to ask e.g.
"which executables do you ship in this gem?"
All this goes away now. We load the gem file in the script and can
work based on the informations we find in the gem spec.
While fixing this, I noticed we have way to many places where you have
to find all versioned ruby interpreters, gem or gem2rpm do something
with them. for this we have a small shell script now.
/usr/bin/ruby-find-versioned [program]
If nothing is passed as program, it will just return all ruby
interpreter. Otherwise the versioned files found for the program.
All the macros in ruby-common have be fixed already. you can check any
post_install scripts that you might have and make use of the little
shell script.
One minor fix you might have to do now on your packages:
If you see your package fail with an error
that /usr/share/doc/packages/$ruby-rubygem-$gemname is not found, it
might be you have to regenerate the spec file to pick up the
addition of .markdown to the list of allowed documentation extensions.
Hopefully in the long run we can move the automation of this also into
gem_install.sh and out of the main spec file template.
With kind regards,
Marcus Rueckert
[1] Yes this really caused bugs.
--
openSUSE - SUSE Linux is my linux
openSUSE is good for you
www.opensuse.org
--
To unsubscribe, e-mail: [email protected]
To contact the owner, e-mail: [email protected]
