-
Notifications
You must be signed in to change notification settings - Fork 262
Conversation
go.mod contains many replace entries that all consumers must include to import logic. These are cancerous and tend to make consumers builds difficult or outright fail. All have been removed, except for the one required by helm-operator. This entry, while retained, has been converted into a noop and should be removed when fluxcd/helm-operator#97 is fixed.
cb9ef8c
to
30a19cd
Compare
d9c1d88
to
c3a879c
Compare
go.mod
Outdated
// TODO(carnot-snap): force downgrade because of a trasitive upgrade | ||
// | ||
// github.com/fluxcd/helm-operator | ||
// +-> github.com/fluxcd/flux@v1.15.0 | ||
// +-> github.com/fluxcd/helm-operator@v1.0.0-rc1 | ||
// +-> github.com/weaveworks/flux@v0.0.0-20190729133003-c78ccd3706b5 | ||
// +-> k8s.io/code-generator@v0.0.0-20190511023357-639c964206c2 | ||
// | ||
// This replace should be removed once fluxcd/flux is released. | ||
replace k8s.io/code-generator => k8s.io/code-generator v0.0.0-20190311093542-50b561225d70 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
removal of this dependency cycle will require synchronised commits in both helm-operator
and flux
:
- remove all
require
andreplace
blocks from helm-operator: carnott-snap/helm-operator@100d4a6 - pin
flux
to pseudo version ofhelm-operator
: carnott-snap/flux@65da8ef - remove unnecessary
replace
blocks:
c3a879c
to
9d513a0
Compare
When you say this, what would you like to see teased apart? I think you are referring to the migration from
My |
Yes, I am referring to the renames. They don't belong to this PR. I am still unsure of whether we should merge them but, if anything, they should be placed in a separate PR. |
I have isolated and reduced that change in #101, and will wait till that merges before we proceed. Are you interested in restarting the dependency cycle as described above, so that we can cleanup the If we cannot come to an understanding on removing them, I would be interested to fixup the |
9d513a0
to
e768511
Compare
@carnott-snap as you may have noticed I am refactoring large parts (if not all) of the code this touches in #99. Would you be open to make this PR target the It is not likely that there will be a release anyway before those two have been merged together, so release wise it does not make a difference. |
I am happy to make the change against |
go.mod contains many replace entries that all consumers must include to import logic. These are cancerous and tend to make consumers builds difficult or outright fail. All unnecessary replaces have been removed. One is required by flux and should be removed when distribution/distribution#2905 is released. The second is caused by a co-dependant transitive upgrade that bumps k8s.io/code-generator beyond kubernetes-1.14.4. To correct this the go.mod has been left untidy, with the correct version required and replaced so that build/test will pass.
e768511
to
85909d9
Compare
@hiddeco, after further review, it appears that This issue is caused by your
|
@carnott-snap can you revise this PR so that it is mergeable with master, and write down the steps that you have in mind that should happen after this has been merged? As it is hard for me to determine what has and has not been done at this point. Thanks 🥇 |
go.mod contains many replace entries that all consumers must include to
import logic. These are cancerous and tend to make consumers builds
difficult or outright fail. All have been removed, except for the one
required by flux. This entry, while retained, has converted into a noop
and should be removed when fluxcd/flux#2590 is fixed.