Updating bundler in EPEL6

Mo Morsi mmorsi at redhat.com
Tue Apr 1 14:36:07 UTC 2014


On 03/27/2014 04:49 PM, Mo Morsi wrote:
> Greetings, there are a couple of bugs [1] [2] filed against the outdated
> version of rubygem-bundler in EPEL.
>
> Looking at the bundler changelog [3], I do not see any noticeable
> regressions / deprecations since the version we currently ship via el6
> (1.1.4) so this update should be fine w/ the EPEL update policy [4].
>
> I've updated my local spec to the version in rawhide (1.5.2) and have
> run a working scratch build against the el6-candidate [5].
>
> Are there any objections to proceeding with this? If not I'll run the
> update sometime next week.
>
> Take care,
>   -Mo
>
> [1] https://bugzilla.redhat.com/show_bug.cgi?id=908061
> [2] https://bugzilla.redhat.com/show_bug.cgi?id=1078539
> [3] https://github.com/bundler/bundler/blob/master/CHANGELOG.md
> [4] https://fedoraproject.org/wiki/EPEL_Updates_Policy
> [5] http://koji.fedoraproject.org/koji/taskinfo?taskID=6680738

Updated bundler, ran the build [1], and submitted it [2]

Had to downgrade ruby dep to 1.8 as that is what is available in the
build environment. Previous build (pre-update) was not working [3]

This introduces an interesting situation, while we can relax the
'Requires: ruby(abi)' version in the spec, only the older version of the
ruby library defines this macro, so ruby 1.8 will be pulled in
regardless. And it can't be changed to 'ruby(release)' as the older ruby
does not define this macro.

Would be nice to be able to update this so that if admins get a newer
ruby rpm from another source, they can just use that in lieu of also
pulling in ruby 1.8 to satisfy this dep. Any thoughts on the feasibility
of this? Perhaps the newer ruby macros can also define 'abi' for
compatibility purposes? Or something else all together?

  -Mo

[1] http://koji.fedoraproject.org/koji/taskinfo?taskID=6696082
[2] https://admin.fedoraproject.org/updates/rubygem-bundler-1.5.2-2.el6
[2] http://koji.fedoraproject.org/koji/taskinfo?taskID=6696062


More information about the ruby-sig mailing list