We're happy to announce the first Encore release following the open source announcement two weeks ago: v0.13!
The headline feature of Encore v0.13 is built-in support for Structured API Errors. Beyond that we've made a lot of minor adjustments to improve the onboarding experience, and fixed a lot of bugs that you've reported!
To upgrade, run the appropriate command:
brew install encoredev/tap/encore # macOS
iwr https://encore.dev/install.ps1 -useb | iex # Windows
curl -L https://encore.dev/install.sh | bash # Linux
We're also happy to announce we've started doing Office Hours!
If you want to chat about anything Encore-related, book a session with us here! We love hearing about your experiences with Encore, and what we can do to make it even better.
With that, let's dig in to the release highlight!
Encore now supports returning structured error information from your APIs.
The key piece is the new encore.dev/beta/errs package.
Structured errors are represented by the errs.Error
type:
type Error struct {
// Code is the error code to return.
Code ErrCode `json:"code"`
// Message is a descriptive message of the error.
Message string `json:"message"`
// Details are user-defined additional details.
Details ErrDetails `json:"details"`
// Meta are arbitrary key-value pairs for use within
// the Encore application. They are not exposed to external clients.
Meta Metadata `json:"-"`
}
Returning an *errs.Error
from an Encore API endpoint will result in Encore
serializing this struct to JSON and returning it in the response. Additionally
Encore will set the HTTP status code to match the error code (see the mapping table below).
For example:
return &errs.Error{
Code: errs.NotFound,
Message: "sprocket not found",
}
Causes Encore to respond with a HTTP 404
error with body:
{
"code": "not_found",
"message": "sprocket not found",
"details": null
}
Encore applications are encouraged to always use the errs
package to
manipulate errors. It supports wrapping errors to gradually add more error
information, and lets you easily define both structured error details to return
to external clients, as well as internal, key-value metadata for debugging
and error handling.
func Wrap(err error, msg string, metaPairs ...interface{}) error
Use errs.Wrap
to conveniently wrap an error, adding additional context and converting it to an *errs.Error
.
If err
is nil it returns nil
. If err
is already an *errs.Error
it copies the Code, Details, and Meta fields over.
The variadic metaPairs
parameter must be key-value pairs, where the key is always a string
and the value can be
any built-in type. Existing key-value pairs from the err
are merged into the new *Error
.
func WrapCode(err error, code ErrCode, msg string, metaPairs ...interface{}) error
errs.WrapCode
is like errs.Wrap
but also sets the error code.
func Convert(err error) error
errs.Convert
converts an error to an *errs.Error
. If the error is already an *errs.Error
it returns it unmodified.
If err
is nil it returns nil.
The errs
package defines error codes for common error scenarios.
They are identical to the codes defined by gRPC
for interoperability.
The table below summarizes the error codes. You can find additional documentation about when to use them in the package documentation.
Code | String | HTTP Status |
---|---|---|
OK | "ok" | 200 OK |
Canceled | "canceled" | 499 Client Closed Request |
Unknown | "unknown" | 500 Internal Server Error |
InvalidArgument | "invalid_argument" | 400 Bad Request |
DeadlineExceeded | "deadline_exceeded" | 504 Gateway Timeout |
NotFound | "not_found" | 404 Not Found |
AlreadyExists | "already_exists" | 409 Conflict |
PermissionDenied | "permission_denied" | 403 Forbidden |
ResourceExhausted | "resource_exhausted" | 429 Too Many Requests |
FailedPrecondition | "failed_precondition" | 400 Bad Request |
Aborted | "aborted" | 409 Conflict |
OutOfRange | "out_of_range" | 400 Bad Request |
Unimplemented | "unimplemented" | 501 Not Implemented |
Internal | "internal" | 500 Internal Server Error |
Unavailable | "unavailable" | 503 Unavailable |
DataLoss | "data_loss" | 500 Internal Server Error |
Unauthenticated | "unauthenticated" | 401 Unauthorized |
In cases where you have complex business logic, or multiple error returns, it's convenient to gradually add metadata to your error.
For this purpose Encore provides errs.Builder
. The builder lets you
gradually set aspects of the error, using a chaining API design.
Use errs.B()
to get a new builder that you can start chaining with directly.
When you want to return the constructed error call the .Err()
method.
For example:
func getBoard(ctx context.Context, boardID int64) (*Board, error) {
// Construct a new error builder with errs.B()
eb := errs.B().Meta("board_id", params.ID)
b := &Board{ID: params.ID}
err := sqldb.QueryRow(ctx, `
SELECT name, created
FROM board
WHERE id = $1
`, params.ID).Scan(&b.Name, &b.Created)
if errors.Is(err, sqldb.ErrNoRows) {
// Return a "board not found" error with code == NotFound
return nil, eb.Code(errs.NotFound).Msg("board not found").Err()
} else if err != nil {
// Return a general error
return nil, eb.Cause(err).Msg("could not get board").Err()
}
// ...
}
When you call another API within Encore, the returned errors are always wrapped in *errs.Error
.
You can inspect the error information either by casting to *errs.Error
, or using the below
helper methods.
func Code(err error) ErrCode
errs.Code
returns the error code. If the error was not an *errs.Error
it returns errs.Unknown
.
func Meta(err error) Metadata
type Metadata map[string]interface{}
errs.Meta
returns any structured metadata present in the error. If the error was not an *errs.Error
it returns nil.
Unlike when you return error information to external clients,
all the metadata is sent to the calling service, making debugging even easier.
func Details(err error) ErrDetails
errs.Details
returns the structured error details. If the error was not an *errs.Error
or the error lacked details,
it returns nil.
We've lots of great things planned for Encore.
The next big thing we're working on is being able to integrate Encore applications with existing infrastructure, to more easily adopt it for existing applications.
This means being able to connect it up to existing databases, being able to deploy to existing Kubernetes clusters, and even deploying Encore applications to regular VMs and other deployment options. We'll start on the database side and gradually expand our support in this space.
We're also looking at expanding our database support, adding a GraphQL API, better support for ORMs for people who prefer that, and more.
We're still working on gathering all your feedback and assembling it into a roadmap of some kind, but please continue letting us know what you'd like to see!
The best way is to open an issue at github.com/encoredev/encore, or reach me on Twitter.