Wrong Response Request Body about govcr HOT 3 CLOSED

Hey, that's a nifty find!

It's been on my radar that some of these helper methods aren't directly tested...

Impact is a little indirect i.e. when Response.Request.Body is referenced. I expect this to be uncommon but nonetheless that's wrong as it stands right now. It does not affect request matching (which obviously takes place before the response is provided).

The contents of the cassette on file is not affected because the whole Response.Request is constructed after the cassette is loaded and after the request has been matched. Note that in the HTTP transaction itself, Response.Request.Body is always nil as per https://pkg.go.dev/net/http#Response Request's Body is nil (having already been consumed) and govcr respects this.

The only time Response.Request.Body may be populated is transiently for the request mutators.
GoVCR ref:

The track replaying mutator additionally receives an informational copy of the current HTTP request in the track's Response under the Request field i.e. Track.Response.Request. This is useful for tailoring track replays with current request information. See TestExample3 for illustration.

This happens in pcb.replayTrack().

Again, real cool find. I'll patch this first and then beef up the tests for future-proofing 🤦

from govcr.

v12.1.0 fixes the issue.
I'll keep this issue open until I've added the tests.

from govcr.

Tests added.

Thanks for raising this issue 👍

from govcr.