I came across this when parallelizing some backend HTTP calls.
The short of it is that NewStringResponder(), NewBytesResponder(), etc... are all subject to racing since ResponderFromResponse() will create new http.Response objects use the same underlying Body object in each response.
My assumption would be that supplying a string in NewStringResponder() would be free of races. If so - I believe a possible fix is to inline ResponderFromResponse() into each method and ensure NewRespBodyFromString() or NewRespBodyFromBytes() is called for each new http.Response object created.
I came across this when parallelizing some backend HTTP calls.
The short of it is that NewStringResponder(), NewBytesResponder(), etc... are all subject to racing since ResponderFromResponse() will create new http.Response objects use the same underlying Body object in each response.
My assumption would be that supplying a string in NewStringResponder() would be free of races. If so - I believe a possible fix is to inline ResponderFromResponse() into each method and ensure NewRespBodyFromString() or NewRespBodyFromBytes() is called for each new http.Response object created.