RFC: Relax Mixin Contract #109
Conversation
Signed-off-by: Stephen Levine <[email protected]>
Signed-off-by: Stephen Levine <[email protected]> Co-authored-by: Terence Lee <[email protected]>
Signed-off-by: Stephen Levine <[email protected]>
|
Should be ready for votes again. |
|
Final Comment Period with merge disposition, closing on 17 September, 2020. |
| # Drawbacks | ||
| [drawbacks]: #drawbacks | ||
|
|
||
| This change reduces the power of the mixin model for stack authors. Previously, stack authors could reliably use `run:mypkg` to indicate that a run image strictly contains the runtime version of mypkg (which may provide different functionality than `mypkg`). Now `run:mypkg` and `mypkg` imply the exact same thing for the run image. |
There was a problem hiding this comment.
Would you mind clarifying this point? Are run:mypkg and mypkg different packages? If yes, why do they satisfy each other, and if not, why would they provide different functionality? A concrete example would be fantastic!
There was a problem hiding this comment.
I believe they could be different packages. The apt contract we have applies only to the io.buildpacks.stacks.bionic stack as per this RFC.
There was a problem hiding this comment.
The set of changes to the run image represented by run:mypkg and mypkg are the same, so if mypkg represents an OS package in the context of the run image, run:mypkg must represent that package as well.
[#109] Signed-off-by: Emily Casey <[email protected]>
Readable.