Non-tokio runtime compatibility #46
Conversation
| if status.is_success() { | ||
| Ok(response.body_json().await?) | ||
| } else { | ||
| Err(Error::HttpError(StatusCode::from(u16::from(status)))) |
There was a problem hiding this comment.
Does surf use http crate status code? If yes I think we should drop the httpstatus dependency as well
There was a problem hiding this comment.
Surf uses http_types.
This would also create additional backwards incompatibility since it leaks out too in case that matters. Same group that created surf.
There was a problem hiding this comment.
Hmm, I think we should take the opportunity of making a breaking change and think about other good changes. My understanding is that http is the most standard source of basic http types. httpstatus is definitely not it. Do you think http_types is a good idea to be the type we export out of this crate's error type?
There was a problem hiding this comment.
On the topic of making breaking changes, scryfall itself makes breaking changes all the time, so this crate has to make a lot of them as well. To be precise, wizards makes a lot of breaking changes 🤣
There was a problem hiding this comment.
I can't say I have a ton of experience with it. I think it is better quality than httpstatus but haven't done a thorough comparison either. From a quick survey, StatusCode in http-types and http looks pretty similar.
Options:
http-types- zero impedance mismatch withsurfhttp- most popular interfaceu16- Remove the public facing dep entirely as just return au16, let the client decide if they want to "interpret" it by pulling it into their favoriteStatusCodeobject. Prevents consumers from being pinned to a specific version of the library too.
I actually think the last option might be the best.
There was a problem hiding this comment.
I think we should go with the last option them. Makes sense to me
|
Thanks for this PR, TIL about surf! For my own learning, I'm curious about your motivation for this PR, I myself only use Tokio, tried async-std once but never really used, can you tell me some reasons for preferring async-std over Tokio? |
|
@mendess The main motivation is to use this within a bevy app. It is not directly compatible with Tokio but uses it's own internal executor that is compatible with standard futures. Surf doesn't specifically rely on tokio or async-std but at least enables compatibility with either, in addition to any other executor that uses standard futures (such as in the case of bevy). I'm no expert in this but this is what I have concluded from my time boxed investigation. |
|
We could also make a larger refactoring to the API and add a |
I really like this idea. This was my first real rust project so it naturally misses some good ideas haha. Feel free to make a PR with this change! And thanks for the idea! |
|
I will see if I have the time and interest to work on it. No promises. |
Why?
Make scryfall-rs usable under non-tokio async runtimes.
Reqwest is only compatible with tokio. Surf is compatible with async-std, tokio, or any other standard futures compliant runtime.
seanmonstar/reqwest#719 (comment)
Known Issues
This will break backwards compatibility in the API. Specifically in regards to the removal of the
ReqwestErrorvariant.It is not abstracted in
HttpClientErroras astringso going forward, any changes to the http client shouldn't have an impact on the public API.