FWIW, I would leave the shared folder setup as-is. I use this vagrant setup 
regularly for other projects and I’ve never noticed a performance penalty - 
it’s always felt as responsive as accessing local files. I would think people 
kind of expect this setup when using vagrant (if they’ve used it before 
anyway). And if we change it, we’ll only end up replacing the `git clone` 
instructions with nfs mount or rsync instructions.  


On Wednesday, April 2, 2014 at 6:40 PM, Dave Brondsema wrote:

> I'm working on building a new vagrant image, since the last image we built was
> back in August. Our instructions [1] specify doing a git clone of the repo 
> (and
> I actually missed that and got tripped up by lack of source!). But would it
> make more sense to include the git repo within the machine image? That'd make
> it easier to get up and running (especially if you don't have git installed on
> the host). Code getting stale wouldn't be an issue since our Vagrant
> instructions specify to run update.sh (http://update.sh) which includes a git 
> pull.
>  
> The downside is that the allura source wouldn't be in a shared vagrant folder
> for easy access with your favorite editor, it'd have to be edited with vim
> inside the vagrant box. But shared vagrant folders are pretty slow anyway, so
> its probably better to recommend development via a different shared method
> (vagrant supports NFS or rsync now). See speed comparison at [2].
>  
> A different option would be to recommend downloading an official release 
> instead
> of git clone.
>  
> Yes? No?
>  
> [1]
> https://forge-allura.apache.org/p/allura/wiki/Install%20and%20Run%20Allura%20-%20Vagrant/
> [2] http://mitchellh.com/comparing-filesystem-performance-in-virtual-machines
>  
> --  
> Dave Brondsema : [email protected] (mailto:[email protected])
> http://www.brondsema.net : personal
> http://www.splike.com : programming
> <><
>  
>  


Reply via email to