Fix bugs on Watcher and connect()/close(). #19
Conversation
Fix Connect and Watch iterator
Upgrade to proto 4
|
Hey @jpark-prestolabs , many thanks for the changes and continous updates. Unfortunately, I didn't have enough time to review the MR. I'll try to look at it over the weekend. |
|
@jpark-prestolabs if you don't mind I'll split the MR into several chunks and cherry-pick by one. |
Thank you for your review. It's good to me. |
| NOTE: It could be called while other operation is being executed. | ||
| At that time, the executing operation will be broken. | ||
| """ | ||
| for _ in range(MAX_COUNT_TO_WAIT_CONNECT_CHECK): |
There was a problem hiding this comment.
If we already have _connection_check_timeout what is the reasoning behind MAX_COUNT_TO_WAIT_CONNECT_CHECK and multiple tries?
There was a problem hiding this comment.
It's long time ago for me to update it. Sorry I cannot remember the exact reason.
Probably _connection_check_timeout could be enough.
There was a problem hiding this comment.
Now I remember that it is for etcd cluster.
MAX_COUNT_TO_WAIT_CONNECT_CHECK is used for making a connection.
One host name is used for multiple etcd instances in a cluster. So I'd like to make multiple trial.
But here, it could not be required.
There was a problem hiding this comment.
I've created an issue #20 to natively support multiple endpoints.
|
I will mostly retain the old formatting as it is highly depends on tooling. The only way to fix it consistently is to intergrate black into the pipeline. |
| @@ -203,6 +271,20 @@ async def __aenter__(self): | |||
| async def __aexit__(self, *args): | |||
| await self.close() | |||
|
|
|||
| @_handle_errors | |||
| @_ensure_connected | |||
| async def reset_auth_token(self) -> None: | |||
There was a problem hiding this comment.
This probably should be refactored into private _auth and public auth and reused in connect.
|
@jpark-prestolabs I cherry-picked everything except |
This fix includes,