Exactly. Plus some other issues we wanted to solve with the new cache
format.

Apparently though, the old cache format had its charms as well :) Especially
for users interested in poking around in the cache.

Cheers!

On Wed, Aug 3, 2011 at 8:28 AM, Johan Stuyts <[email protected]> wrote:

> Hi,
>
> From my experience dependency managers, by default, use a shared cache for
> all repositories. This results in cache contamination: projects which cannot
> find a dependency using an empty cache are able to find a dependency because
> it was downloaded for another project with different configured
> repositories. So when someone adds a dependency which cannot be found in the
> repositories configured for the project, but that can be found in the cache,
> and checks it in, the build will break on machines where that dependency is
> not already in the cache.
>
> In M3 this was also the case, but with M4 this has been resolved due to the
> new cache structure. I was wondering if the prevention of cache
> contamination was a requirement that resulted in the new cache structure, or
> whether it was just a coincidental result and the new cache structure is
> based on other requirements.
>
> If cache contamination prevention is by design, that would be great because
> then I don't have to enforce a project-specific user home directory for each
> project/branch in my build script wrapper. And it would prevent downloading
> artifacts from repositories used by multiple projects multiple times.
>
> --
> Regards, Johan
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>    http://xircles.codehaus.org/manage_email
>
>
>


-- 
Szczepan Faber
Principal engineer@gradleware
Lead@mockito

Reply via email to