fix: make "other" error actually transparent #51
Conversation
|
You need to run cargo fmt. Would also be worth asking the displaydoc upstream about fixing transparent errors. |
| @@ -166,7 +166,8 @@ async fn sign_and_send( | |||
| task.private_key.clone(), | |||
| task.http_signature_compat, | |||
| ) | |||
| .await?; | |||
| .await | |||
| .context("signing request")?; | |||
There was a problem hiding this comment.
Instead of these context it seems much cleaner if you add new variants to the error enum.
There was a problem hiding this comment.
I don't think it makes much sense to pass all errors as structured errors to the outside - e.g. in this case the private key is broken which is kinda something that has to be fixed manually anyways. But it's really useful to at least get a tiny bit of context for what happened, which is what these changes do, with just a tiny change instead of a large change
|
I've run cargo fmt.
I think there is already a old open issue about this: yaahc/displaydoc#15 |
|
Alright thanks! |
This PR fixes the Display trait for the "Other" error. Before, these errors would just output "Other errors which are not explicitly handled" for everything without any other context.
with this change, they output the full anyhow content.
e.g.: "data did not match any variant of untagged enum PersonOrGroup"
or: "Request error: error sending request for url (http://localhost:5313/u/Lady_Albedo_96): connection error: Connection reset by peer (os error 104)"
The reason for this issue was some bad interaction between the "displaydoc" library and the "thiserror" library. I tried a few other things to fix it but nothing except removing DisplayDoc worked.
This PR also adds a few random "context" calls that I needed while debugging something. They aren't great but imo the more anyhow context is used the better.