Skip to content
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

Update APIError struct to use new NullAPIErrorObject type for safety #272

Merged
merged 1 commit into from Feb 13, 2021

Conversation

theckman
Copy link
Collaborator

I discovered that my initial implementation of the APIError type's NotFound()
method had the potential for triggering a nil pointer dereference panic, which
wouldn't be a great experience for someone dealing with PagerDuty code.

I provided a fix in #271, but in doing so I remembered this pattern from the
database/sql package and decided that, while not as nice to use, it was a much
safer API to reduce the chances of the consumer accidentally triggering a panic
in their program.

Since we've yet to release v1.4.0 (i.e., the new APIError type still hasn't been
released) we can still change this without breaking API compatibility
guarantees.

@theckman
Copy link
Collaborator Author

@echlebek I could use your eyes on this one too, since it was your issue that spawned this API.

@theckman theckman force-pushed the safer_api_error_object_handling branch 3 times, most recently from f85a8d8 to 168142c Compare February 12, 2021 10:53
I discovered that my initial implementation of the APIError type's NotFound()
method had the potential for triggering a nil pointer dereference panic, which
wouldn't be a great experience for someone dealing with PagerDuty code.

I provided a fix in #271, but in doing so I remembered this pattern from the
`database/sql` package and decided that, while not as nice to use, it was a much
safer API to reduce the chances of the consumer accidentally triggering a panic
in their program.

Since we've yet to release v1.4.0 (i.e., the new APIError type still hasn't been
released) we can still change this without breaking API compatibility
guarantees.
@theckman theckman force-pushed the safer_api_error_object_handling branch from 168142c to 5e16250 Compare February 12, 2021 10:54
Copy link

@echlebek echlebek left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice catch. It does improve the safety of the library to remove the chance of a nil pointer deref.

In terms of ergonomics, I wonder if it wouldn't be a little nicer to make methods on APIError instead? For example:

type APIError struct {
  statusCode int
  apiError *APIErrorObject
}

func (a APIError) StatusCode() int {
  return a.statusCode
}

func (a APIError) ErrorCode() int {
  if a.apiError != nil {
    return a.apiError.Code
  }
  return 0
}

func (a APIError) Message() string { ... }

func (a APIError) Errors() []string { ... }

@theckman
Copy link
Collaborator Author

theckman commented Feb 12, 2021

Thank you for your thoughtful reply, I honestly considered it doing that way first and eventually talked myself out of it.

The biggest concern I had is that we then still need to have a Valid() method to know that the data in the methods are valid, or you have to use the comma ok idiom to let callers know that it's a good value to use. Unfortunately, we can't use the same trick maps or channels use and so consumers wouldn't be able to do things like fmt.Printf("Status Code: %d\n", aerr.StatusCode()) if we did that.

I also know people work with the sql.Null* types quite often, and so the ergonomics won't be foreign
to many folks. So, considering that I decided to go with the structure you see here.

Copy link
Contributor

@stmcallister stmcallister left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good call. Do you want me to wait to cut a new release?

@theckman theckman merged commit 1df5787 into master Feb 13, 2021
@theckman theckman deleted the safer_api_error_object_handling branch February 13, 2021 01:29
@theckman
Copy link
Collaborator Author

@stmcallister alright; merged it in. Thank you for that approval.

I think I'm ready for a v1.4.0 now.

@theckman
Copy link
Collaborator Author

@stmcallister I can look to prep it if you'd like. It seems like I need to get a PR up for the CHANGELOG.md file, but I would need your approval to merge it. LMK if that sounds good.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants