formdata
formdata copied to clipboard
mikedilger
bump hyper and mime vers
I see that this requires also your changes to the mime-multipart crate. I'll start there.
Don't expect a quick turnaround as I'm pretty busy tonight and tomorrow is my Monday which is always busy.
Understood. I'll start work on tests. Thank you!
On Sun, Sep 10, 2017 at 3:12 AM, Michael Dilger [email protected] wrote:
I see that this requires also your changes to the mime-multipart crate. I'll start there.
Don't expect a quick turnaround as I'm pretty busy tonight and tomorrow is my Monday which is always busy.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mikedilger/formdata/pull/37#issuecomment-328324624, or mute the thread https://github.com/notifications/unsubscribe-auth/APKSuAY1LZbG2nAgq0x46663_q7xxbXXks5sg4vUgaJpZM4PSMx7 .
hyper 11 is really quite a different paradigm. It leaves me thinking that the public interfaces of mime-multipart and formdata are not appropriate in the tokio-based world... not that they couldn't work (your forks show that), but that when used this way you lose the asynchronicity (e.g. read_formdata() will block and your asynchronous server will not progress until read_formdata() has completed).
Personally I'm leaning towards entirely new crates, perhaps called mime-mulitpart-async and formdata-async. I think hyper should have done that too.. so that the synchronous code could continue to be developed.
Have you thought about whether you're going to employ nightly features?
On Sep 11, 2017 5:25 PM, "Michael Dilger" [email protected] wrote:
hyper 11 is really quite a different paradigm. It leaves me thinking that the public interfaces of mime-multipart and formdata are not appropriate in the tokio-based world... not that they couldn't work (your forks show that), but that when used this way you lose the asynchronicity (e.g. read_formdata() will block and your asynchronous server will not progress until read_formdata() has completed).
Personally I'm leaning towards entirely new crates, perhaps called mime-mulitpart-async and formdata-async. I think hyper should have done that too.. so that the synchronous code could continue to be developed.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/mikedilger/formdata/pull/37#issuecomment-328663261, or mute the thread https://github.com/notifications/unsubscribe-auth/APKSuL7nSGifLdvk5rf6ySDdq99gn8FMks5shaVYgaJpZM4PSMx7 .