feat: resolve multiple function/event selectors in one openchain.xyz request #6719
Conversation
| } | ||
|
|
||
| /// Identifies `Function` from its cache or `https://api.openchain.xyz` | ||
| pub async fn identify_function(&mut self, identifier: &[u8]) -> Option<Function> { | ||
| self.ensure_not_offline()?; |
There was a problem hiding this comment.
We want to use results from loaded disk cache, network requests will be ignored in identify() function
| } | ||
|
|
||
| #[derive(Deserialize)] | ||
| struct ApiResult { | ||
| event: HashMap<String, Vec<Decoded>>, | ||
| function: HashMap<String, Vec<Decoded>>, | ||
| event: HashMap<String, Option<Vec<Decoded>>>, |
There was a problem hiding this comment.
api returns null for unknown selectors, we don't want to fail whole request with probably success responses
| #[derive(Deserialize)] | ||
| struct Decoded { | ||
| name: String, | ||
| filtered: bool, |
There was a problem hiding this comment.
filter=true by default (source), so results are already filtered on API side and this field is not in response.
|
hey @cdump — def no more big changes planned for this section of code, sorry about that! feel free to go ahead and rebase |
|
Rebased & ready for review. |
|
@Evalir, what do you mean by 'needs clippy'? It seems that other MR no longer fail on clippy nightly. Unfortunately, I don't have the rights to restart the clippy check in this MR, should I force-push just to restart pipeline? |
Motivation
Currently, the number of HTTP requests to openchain.xyz is equal to the number of selectors to resolve, which is not optimal.
On the current
master, the followingcast runtakes ~11 seconds:I've added
println!("get {url}");to selectors.rs:58 and got:Solution
openchain.xyz's api supports multiple selectors in one request - let's use that feature.
After the proposed changes,
cast run ...took 2.8 seconds - x4 faster than before