GitLab

In crrev.com/c/1256016, we moved FidoAuthenticator instantiation and
ownership into the respective FidoDiscovery subclasses. This also
eliminated FidoRequestHandlerBase::CreateAuthenticatorFromDevice() which
was where we previously wrapped a discovered FidoDevice in a
FidoDeviceAuthenticator before adding it to the request handler.

In fido_request_handler_unittest, this was used as a testing seam to
inject a FakeFidoAuthenticator implementation. Injection of the fake was
dropped, but the "real" FidoDeviceAuthenticator was still casted into a
pointer to the fake elsewhere. While this seems to not bother compilers
too much, it's probably a Bad Idea. This CL therefore entirely removes
the FakeFidoAuthenticator class (without changing semantics of the
fixture or the tests).

Bug: 903294
Change-Id: I7a6bad330b1152dc455ded5b6db3d3309f4eb3be
Reviewed-on: https://chromium-review.googlesource.com/c/1340903Reviewed-by: Jun Choi <hongjunchoi@chromium.org>
Commit-Queue: Jun Choi <hongjunchoi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#609536}