| 1 |
On Wed, Jun 8, 2016 at 9:16 AM, Alexander Berntsen <bernalex@g.o> wrote: |
| 2 |
> |
| 3 |
> This could lead to a future where the Gentoo tree is largely |
| 4 |
> superseded. Every user would just have their own repository, where |
| 5 |
> they could pick and choose packages from other users. The Gentoo tree |
| 6 |
> would just focus on a high-quality repository of the basic/core things |
| 7 |
> that everybody needs. Gentoo devs would spend most of their time |
| 8 |
> maintaining curated small and useful repositories. |
| 9 |
> |
| 10 |
|
| 11 |
Do you propose that you can have cross-repo dependencies? |
| 12 |
|
| 13 |
If so that creates a lot of potential issues, even if you do it the NixOS way. |
| 14 |
|
| 15 |
Suppose you have 10 packages, and they each depend on zlib from a |
| 16 |
different repository? If they collide, that is one problem to solve. |
| 17 |
If they don't collide then you have 10 copies of zlib now, and good |
| 18 |
luck making sure they're all secure, and of course now you're |
| 19 |
multiplying the number of "shared" objects you keep in RAM. |
| 20 |
|
| 21 |
If you only allow dependencies back to the main repository then that |
| 22 |
is a pretty big limitation, and I don't see there being any kind of |
| 23 |
fundamental transformation in how Gentoo operates that way. |
| 24 |
|
| 25 |
Don't get me wrong - I'm all for distributed repositories. I just |
| 26 |
think there are some big challenges if we want to have them largely |
| 27 |
supersede the main repo. |
| 28 |
|
| 29 |
I do think that leveraging git makes a lot of sense if your goal is to |
| 30 |
facilitate movement into the main tree. It also has benefits even if |
| 31 |
the repos are all stand-alone, since you at least have unique hashes |
| 32 |
and such to refer to from a version control standpoint, vs just files |
| 33 |
stored in a random directory tree somewhere with no way to tell if any |
| 34 |
have changed and how. |
| 35 |
|
| 36 |
-- |
| 37 |
Rich |
Archives