On Sat, Apr 18, 2009 at 10:02, Darren Boyd <[email protected]> wrote:
> On the topic of putting all gems in a Rails project into vendor, what > do people do with gems that have a native compilation step? We've > been leaving these out of the project source control. Instead, we > install them on each machine as required. Along the lines of Rob's reply, I've thought at times of adding "rake gems:install" to my deploy script. Haven't actually gotten around to it yet (new gems being infrequent enough that cap invoke isn't too big an annoyance) but it may be worth a try. It may be able to offer the advantage of only installing your gems once, but I'm not entirely sure offhand how gems:install works with gems that are already frozen into your app. > And what happens with gems like rmagick, which have a > native part that is tightly coupled to other native libraries? Eh, I just build these out manually. There are some capistrano tasks out there to handle this kind of thing (the deprec gem comes to mind). But, personally, I don't keep these libraries up to the bleeding edge anyway. Plus a lot of my apps are on EC2 now, and AMIs can definitely help mitigate the cost of infrequent manual configuration. -- Nick Zadrozny --~--~---------~--~----~------------~-------~--~----~ SD Ruby mailing list [email protected] http://groups.google.com/group/sdruby -~----------~----~----~----~------~----~------~--~---
