| 1 |
On Mon, Feb 17, 2014 at 12:09 PM, Sebastian Luther |
| 2 |
<SebastianLuther@×××.de>wrote: |
| 3 |
|
| 4 |
> Am 17.02.2014 18:52, schrieb David James: |
| 5 |
> > @@ -96,6 +105,68 @@ class MultirepoTestCase(TestCase): |
| 6 |
> <snip> |
| 7 |
> > + |
| 8 |
> > + #--newrepo: pick ebuild if binpkg/ebuild have |
| 9 |
> different repo |
| 10 |
> > + ResolverPlaygroundTestCase( |
| 11 |
> > + ["dev-libs/C"], |
| 12 |
> > + options = {"--usepkg": True, "--newrepo": |
| 13 |
> True, "--selective": True}, |
| 14 |
> > + success = True, |
| 15 |
> > + check_repo_names = True, |
| 16 |
> > + mergelist = ["dev-libs/C-1::repo1"]), |
| 17 |
> |
| 18 |
> Why should it take the ebuild from the repo with lower priority? |
| 19 |
> |
| 20 |
> I thought the intend was that the package should be reinstalled if it is |
| 21 |
> now pulled from a repo with higher priority than the repo from which the |
| 22 |
> installed package came. |
| 23 |
|
| 24 |
|
| 25 |
Interesting idea -- but no, that's not what I implemented. The intent here |
| 26 |
is that a package will be reinstalled if it is now pulled from a different |
| 27 |
repo than the repo from which the installed package came. This is actually |
| 28 |
an important difference for our use case -- it allows you to switch in both |
| 29 |
directions, both to higher priority repos and to lower priority repos. |
| 30 |
|
| 31 |
From a theoretical perspective, the reason why I implemented this is |
| 32 |
because "ebuilds" are the source of truth. According to the ebuilds, the |
| 33 |
owner repo of dev-libs/C-1 is repo1, so this means that, with --newrepo, |
| 34 |
any binpkg for dev-libs/C-1 from repo2 would be invalid. |
| 35 |
|
| 36 |
This functionality is useful in the case where an ebuild changes repos. In |
| 37 |
this case, with --newrepo, you would want the package to be rebuilt and any |
| 38 |
binaries to be invalidated, and our tests validate this. |
| 39 |
|
| 40 |
Cheers, |
| 41 |
|
| 42 |
David |
Archives