Implement take_prev_kvs for DeleteResponse #68
Conversation
Signed-off-by: tison <[email protected]>
|
cc @davidli2010 |
|
@davidli2010 Do you think this patch in a good direction? Also, I wonder if we have an estimate for a next release so that downstream can upgrade to a version with new patches >_< |
This PR is OK, but need to improve the implementation according to my comment. There is no regular release plan, just release as needed. |
I don't see extra comments yet. |
src/rpc/kv.rs
Outdated
| /// If `prev_kvs` is set in the request, take the previous key-value pairs, leaving an empty vector in its place. | ||
| #[inline] | ||
| pub fn take_prev_kvs(&mut self) -> Vec<KeyValue> { | ||
| self.0.prev_kvs.drain(..).map(KeyValue::new).collect() |
There was a problem hiding this comment.
No need to collect KeyValue into another vector, as DeleteResponse/KeyValue are just wrappers of PbDeleteResponse/PbKeyValue, here we can take and transmute prev_kvs to Vec:
unsafe { std::mem::transmute(std::mem::take(&mut self.0.prev_kvs)) }Signed-off-by: tison <[email protected]>
|
@davidli2010 Thanks for your review. According to release on demand, perhaps I can expect a release in these days? |
|
v0.12.2 is released |
|
Thank you! |
Just like
take_prev_keyforPutResponse.I found this helper function help when HACKING GreptimeTeam/greptimedb#2734.