(to both cloud@ and devel@)
I was interested in a yum plugin to support S3, and discovered this
forum post, where Seth and James basically gave their thumbs up. The
author (Robert Mela, even though he's not the original author either)
made it work and there was some feedback provided, which I incorporated,
along with a bunch of small improvements in my github branch.
Robert says in the forum post that he'll be happy if someone shepherds
this into completion and I'm more than happy to try.
Right now the plugin works, is a bit more aligned with the yum plugin
architecture I think (I'd really like it to be true :) and also allows
you to set the S3 credentials on a global, per-repo or using environment
There are 2 things I want to complete/need help with:
- Package it as an RPM: I have a half-finished spec for this and I
should have no problems finishing as long as I can deal with the item
- Fix an annoying bug where the plugin creates an empty yum cachedir in
the current directory. This is because the way the plugin works, it
creates a copy of the original repo with a different grabber based on
boto/urllib2 and then copies the rest of the attributes. However when
the new repo is created, basecachedir is empty. Can any of you fine
people spot a way around that? it's been bugging me for a good week now.
thoughts? comments? suggestions? let me know :)
PS if you're asking yourselves: Why would you need this plugin if you
can create websites off an S3 bucket? the answer is: using S3 offers
more granularity to share, instead of just opening up the contents of
the bucket to the entire world.