Prevent breakage when merging OSS changes to ENT due to missing funcs #21564
Conversation
…e stubs on oss, and also on ent if the real func hasn't yet been defined.
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
ncabatoff
changed the title
tools/stubmaker/main.go
Outdated
| } | ||
|
|
||
| target := fmt.Sprintf("%s_ent.go", baseFilename) | ||
| os.Remove(target) |
There was a problem hiding this comment.
do we want to remove the ent file, if it exists?
There was a problem hiding this comment.
I think we do. It's possible that we created it in a previous run, but since then the real funcs were implemented, and now the presence of the file causes builds to fail due to multiple definitions of the funcs.
| cfg := &packages.Config{ | ||
| Mode: packages.NeedName | packages.NeedTypes | packages.NeedTypesInfo | packages.NeedSyntax, | ||
| Tests: false, | ||
| BuildFlags: []string{fmt.Sprintf("-tags=%s", strings.Join(tags, " "))}, |
There was a problem hiding this comment.
do we need to worry about any additional tags, like hsm or fips?
There was a problem hiding this comment.
Good question. Re HSM, I don't think so, we don't run tests using hsm AFAIK, I think it's mostly about how we build the binaries. Re fips, less certain, but looking at how much it's used now I'm inclined to not worry about. For one thing, how would we know which to use when running locally (i.e. not in CI), fips or not fips?
|
I'll defer backporting until I've done more testing. |
See https://hashicorp.atlassian.net/browse/VAULT-17420 for background.