Documentation ¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type UploadReader ¶
type UploadReader struct {
// contains filtered or unexported fields
}
UploadReader is a pass-through reader for use in sending non-trivial data using the net/http package (http.NewRequest, http.Post and the like).
The net/http package uses a separate goroutine to upload data to a HTTP connection, and it is possible for the server to return a response (typically an error) before consuming the full body of the request. In that case http.Client.Do can return with an error while the body is still being read — regardless of of the cancellation, if any, of http.Request.Context().
As a result, any data used/updated by the io.Reader() provided as the request body may be used/updated even after http.Client.Do returns, causing races.
To fix this, UploadReader provides a synchronized Terminate() method, which can block for a not-completely-negligible time (for a duration of the underlying Read()), but guarantees that after Terminate() returns, the underlying reader is never used any more (unlike calling the cancellation callback of context.WithCancel, which returns before any recipients may have reacted to the cancellation).
func NewUploadReader ¶
func NewUploadReader(underlying io.Reader) *UploadReader
NewUploadReader returns an UploadReader for an "underlying" reader.
func (*UploadReader) Read ¶
func (ur *UploadReader) Read(p []byte) (int, error)
Read returns the error set by Terminate, if any, or calls the underlying reader. It is safe to call this from a different goroutine than Terminate.
func (*UploadReader) Terminate ¶
func (ur *UploadReader) Terminate(err error)
Terminate waits for in-progress Read calls, if any, to finish, and ensures that after this function returns, any Read calls will fail with the provided error, and the underlying reader will never be used any more.
It is safe to call this from a different goroutine than Read.