access_loggers: use new-style names

[zuercher]

Modifies the well-known-names of the built-in access loggers to
use the same name as the extension build system.

(This is the first in a series, but I expect whatever changes may be
requested here will apply to all the rest, so I'm going to hold the others
back.)

Risk Level: low, previous name is still accepted
Testing: added test to prove factory accepts old names
Docs Changes: updated names
Release Notes: updated
Deprecated: old names are logged as deprecated

Signed-off-by: Stephan Zuercher [email protected]

[repokitteh-read-only]

CC @envoyproxy/api-shepherds: Your approval is needed for changes made to api/.

🐱

Caused by: #9921 was opened by zuercher.

see: more, trace.

[junr03]

@kyessenov do you mind taking a look?

[kyessenov]

@zuercher After #9618, getAndCheckFactory uses the type URL in the config as the first lookup to pick an extension. That means the name field in the config proto becomes optional. Can you check that is indeed the case by setting it to something random? I haven't tested the access log extensions, only the HTTP extensions. There might be a difference depending how the factory is selected.

[zuercher]

It's correct that with typed_config the name is ignored. Mostly I'm undertaking this change to unify the build extension and config names for v2 in the short term. The top of each protobuf doc currently emits an extension name that isn't necessarily the correct one to use with config.

[kyessenov]

LGTM. I agree we need this to support untyped configs in the short term, for consistency with the rest of the extensions. I can do a follow-up to assign random optional names to extensions later, once we feel that the world has moved on to v3 and 1.13.

[zuercher]

So can we move forward with this?

[mattklein123]

Hero!