-
Notifications
You must be signed in to change notification settings - Fork 3
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
chore: add sdk-test-data as a submodule #22
Conversation
c1fe321
to
26e0726
Compare
26e0726
to
0cd59cb
Compare
Agree this is annoying, however given the sdk-test-data is the source of truth regarding "correct" operation, it seems like its basically an annoying way of notifying that changes have been decided. However, one way or another, we'd want that notification.
I'm not sure this is the plan we want. Would this then make the rust SDK different than others? If we do want to tie the SDK to snapshots of our test data (which I don't think we do), I wonder if just specifying a version hash in the checkout step would be easier |
Yes. I want that notification, too—just want to make it less "annoying." I wouldn't call annoyance as the main reason though—my main concern is that it's breaking too much. It breaks Besides hindering the development flow (having to take a look why CI on my PR or local tests suddenly go bonkers), it also breaks some other less frequently used tools: Besides that, this approach is bad at notifications, too—if an SDK doesn't have new PRs open, there's no notification. Opening a PR is a better way to notify about changes elsewhere (e.g., dependabot is a widely known example).
Hm, yes, Rust (and Ruby 😼) would differ in a small way. If we have a good experience here though, I believe we would want to replicate that behavior to other SDKs.
Not sure I follow. A git submodule is just specifying a version hash in the checkout step. What do you mean? Just to clarify my proposal: when |
I'm probably misunderstanding how submodules work. If basically a change in sdk-test-data results in an auto-created PR for this repository to update to use the latest test data, and leaves all other repositories unaffected, then this does sound good. It sounds conceptually similar to my suggestion, which is to have make-test-data check out a specific version of sdk-test-data, so a PR would be needed to get more recent data. But no automatic PR as in your proposal, which I think we want. |
Submodule is basically a pointer to another repository commit (url + commit hash) that git can pull for you. Seems to be exactly your proposal—just using native git tool. Because it stores a specific hash, we need some way to update it when upstream |
Ok I see, so it would be the combination of the sub module plus an action in the test data repository that automatically opens a PR? Sounds good to try this out here and if we like it, employ it in other repositories. |
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.
Thanks for answering my questions!
I'm somewhat tired of
sdk-test-data
updates breaking this SDK unprompted. I also want the ability to apply test-data fixes in branches (e.g., #13 is currently blocked by Eppo-exp/sdk-test-data#49).Plan:
sdk-test-data
as a submodule so that we can update it when we're ready (instead of updates being pushed on us unexpectedly). (this PR)