POST request with no body results in a rack exception about iodine

[boazsegev]

@thexa4 , thank you very much for exposing this issue. 🎉👏🏻🙏🏻👍🏻

You're right, iodine should have returned an empty string ("") rather than nil. The issue occurred when an optimized "no content" path was taken instead of reaching EOF on the rack.input stream.

I posted an initial patch and I hope to release it soon, after running a few tests.

@SleeplessByte,

I'm not sure what you meant by suggesting to test the content_length?

I'm assuming you were suggesting a way to patch Rack... and although I think your instincts are great, 99.9% of all POST requests probably contain some data. I think it's fair that Rack would upstream the edge-case handling to the server.

[boazsegev]

@SleeplessByte - yes, I agree... the Rack documentation could be improved.

On the other hand, Rack was a huge undertaking and it's so relied upon that it became practically impossible to replace. It's battle tested and dependable. No matter how many of us complain about Rack's CGI core, it's probably here to stay and it's making many developers happy.

[SleeplessByte]

It makes me happy 😉 so it gets my upvote regardless of the 🆗 documentation.

[SleeplessByte]

This should probably also check the content_length before performing any mutation.

[boazsegev]

Patch released. closing issue.

Thanks again! 🙏🏻👍🏻🎉

[SleeplessByte]

@boazsegev you're absolutely right on my intention with that comment, and I agree that this edge case can be handled here, in iodine :). Thank you for the blazingly fast update ❤️

The biggest issue I have with the rack implementation is that the comment for the rack functionality says it's supposed to check if there is data, whereas it's actually checking if there could be data -- and then it's further assumed that there is data.