1

I'm developing a workflow for installing software on Linux that is not available as an RPM, and capturing the system changes to then package as an RPM for internal deployment.

One idea I have for doing this is to leverage btrfs, take a snapshot before, and a snapshot after the install, diff the snapshots, and use that (with some commonsense exclusions) to build an rpm.

Is there a more obvious way of doing this?

mdur
  • 244
  • 2
  • 9
  • What compilation/installation method does the software use? – Etan Reisner Nov 18 '14 at 17:28
  • @EtanReisner It may be anything from a binary that runs in X11 to a shell script. – mdur Nov 18 '14 at 17:49
  • That's not an answer to my question though. The question isn't what the installed software is but what it uses to install itself. All an RPM does is specify an alternate prefix and then package up those files. If you can do that for your software that's all you need. – Etan Reisner Nov 18 '14 at 17:57
  • @EtanReisner In my response I am referring to the installer, not the software installed, and what is used varies between pieces of software. The matlab installer for instance is a binary that is made to be run by a user in X11. – mdur Nov 19 '14 at 19:50
  • I see. The rest of my answer still obtains though. If you can get the installer to run in an automated fashion to a known prefix that's all you need. If you can't do that then something like what you suggested would work. There are also tools like checkinstall which might help you here (though probably not for the matlab case). You could also check for any distributions that package the software and see what they do. – Etan Reisner Nov 19 '14 at 22:42

0 Answers0