Falcon::Adapters::Input#gets doesn't comply with the Rack specification about falcon HOT 4 CLOSED

That makes sense.

from falcon.

We'd need to copy something like https://github.com/socketry/async-io/blob/c39a819204c18864f66f354f1e1034dab31e2874/lib/async/io/stream.rb#L74-L85

from falcon.

The specification for gets is completely insane:

gets(sep=$/ [, getline_args]) → string or nil click to toggle source
gets(limit [, getline_args]) → string or nil
gets(sep, limit [, getline_args]) → string or nil

Reads the next “line'' from the I/O stream; lines are separated by sep. A separator of nil reads the entire contents, and a zero-length separator reads the input a paragraph at a time (two successive newlines in the input separate paragraphs). The stream must be opened for reading or an IOError will be raised. The line read in will be returned and also assigned to $_. Returns nil if called at end of file. If the first argument is an integer, or optional second argument is given, the returning string would not be longer than the given value in bytes.

from falcon.

I started implementing this. It's a huge abomination.

Then, I wondered, what does the rack spec actually say?

https://www.rubydoc.info/github/rack/rack/master/file/SPEC#The_Input_Stream

The input stream is an IO-like object which contains the raw HTTP POST data. When applicable, its external encoding must be “ASCII-8BIT” and it must be opened in binary mode, for Ruby 1.9 compatibility. The input stream must respond to gets, each, read and rewind.

And specifically:

gets must be called without arguments and return a string, or nil on EOF.

It seems like a conforming implementation just needs to call #read, which is exactly what we are doing already.

Please discuss this if you think this interpretation is incorrect, because as I said, it would be good to get this right. But, implementing IO#gets is really nasty.

from falcon.