Improve a bit JournalClient's docstring

"bon d'accord !"

apply vlorentz' suggestions

LGTM.
Now. what you do not understand "I don't understand why..." Is it "why the code does not behaves how it was intended to?" or "why the code was written so it actually uses a local storage?"

LGTM but please fix the abstract class's docstring as stated in the comment above.

update the docstring

rebased

I'm not very fond of testing several things at once. This does not "Test date parsing of origin_metadata_add" as such, it adds this testing as a side effect, in the middle of an unrelated test. Which is not a good idea. I mean, the test is named 'test_origin_metadata_get' and it does already too many things. Having the test named 'test_origin_metadata_get' fail because of bug in parsing dates in origin_metadata_add does not help the poor guy fixing these...

It's not that straightforward that this test does indeed do what the commit message pretends... A comment or 2 in the code of the test would not hurt. But meh

Ok, but I would have loved to see a word or 2 explaining why this revision is needed in the commit message.

log an exception if process_replay_objects_content() fails

rebase

fix the fix

small fix

• Add a content replayer service
• cli: extract the JournalClient instanciation in a dedicated function
• replay: make unexpectedly received object_type log a warning

ok ok, I did split the kafka mock class in 2

add a TODO in test_cli about content-replay

replay: log warnings in case of unexpected object types

In D1541#34664, @vlorentz wrote:

Also, your new CLI needs tests :)

fix an error (brokers can be a tuple) making test_cli fails

The diffs looks fine, but I'd like to have confirmation the code works properly in real conditions (in a docker session) before we merge this.

fix commit msg

should address vlortenz's comments

Several refactorings and git history cleanups

this is now part of D1540

rebase + fix typo