libcontainer: move capabilities to separate package#2607
Conversation
|
While at it, we might start using I am split between the top-level |
|
|
Probably because of I just found myself creating an alternative to this one: #2646 |
|
Ah, yes, avoiding init is definitely good. I still think that having it in a separate package is clean (not critical of course); slightly on the fence on using "internal" (which would hard restrict external users); guess "libcontainer" is now a bit of a confusing name, as it's no longer meant to use as a library (by others) |
|
What's current status? |
|
@kolyshkin good to go? |
|
Could you rebase to retrigger CI? |
e9a3e4b
007ecf7 to
e9a3e4b
Compare
|
rebased 👍 |
|
@kolyshkin @crosbymichael PTAL; this one good to go? |
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
e9a3e4b to
1897217
Compare
|
rebased. @AkihiroSuda @kolyshkin @crosbymichael PTAL; this one good to go? |
|
@cyphar @kolyshkin Can we have this in rc93? |
cyphar
left a comment
There was a problem hiding this comment.
LGTM. Regarding internal I think we should move basically all of libcontainer into internal at some point and move the bits of code people actually use to be outside internal, so that the default state of libcontainer is private.
I had these changes open in the branch I used for #2595. TBH, I don't exactly recall why I made these changes, but I guess because it's slightly cleaner to have it in a separate package (similar to cgroups, devices, apparmor, etc all being separate).
I'm not super-attached to it, so feel free to close if you don't like this change, but thought I'd push it to give the opportunity to review 😅
@kolyshkin @AkihiroSuda