The documentation was correct, the code was not.
This should make it easier to implement new protocols. Note that
for HTTP/2 we will need to add some form of counting later on to
check for malformed requests, but we can do simpler and just
reduce from the expected length and then check if that's 0 when
IsFin=fin.
It is completely removed for connection processes, because
assuming Cowboy is written properly this should bring us
nothing anymore in 2.0.
It is reworked for request processes, there we want to
always propagate the stacktrace (including for exits)
because we will print a report to help with debugging
and proc_lib doesn't propagate it for exits.
At the same time the initial callback for connection
and request processes has been changed to connection_process
and request_process, which should help with identifying
processes when inspecting.
When the request process exits with a {request_error, Reason, Human}
exit reason, Cowboy will return a 400 status code instead of 500.
Cowboy may also return a more specific status code depending on
the error. Currently it may also return 408 or 413.
This should prove to be more solid that looking inside the stack
trace.
I have decided not to include a manual page for
cowboy_stream_h at this point because it clashes
with the cowboy_stream manual page. This decision
will be revisited in the future.
This should work very similar to normal supervisors,
in particular during the shutdown sequence when the
connection process goes down or switches to Websocket.
Processes that need to enforce the shutdown timeout
will be required to trap exits, just like in a supervisor.
In a vanilla Cowboy, this only matters at connection
shutdown, as Cowboy will otherwise wait for the request
process to be down before stopping the stream.
Tests are currently missing.
Cowboy was encoding the headers then decoding them when initializing
the request. The problem is that the encoding and decoding contexts
are not the same. Now, Cowboy will directly use the headers it
received in the push command for the new request. This is also
more efficient.
I am surprised it worked at all considering the issue.
There are two important changes in this commit.
Constraints are now producing an error tuple. This error tuple
in turn can be provided to a function for formatting a human
readable error message. Both the error tuple and the formatting
code are controlled by and part of the constraint function.
Constraints now also implement the reverse operation.
When constraint functions only validate, the reverse operation
will be the same as the forward operation. When they also do
some conversion then the reverse operation will reverse it.
Since constraints are now performing 3 different operations
(forward, reverse and format_error), they now take the form
of a function accepting two separate arguments. The operation
is the first argument.
In addition, the return value was changed to take the form
of {ok, Value} | {error, Reason}. The value must be returned
as-is if it was not modified.
They are now cowboy:start_clear/3 and cowboy:start_tls/3.
The NumAcceptors argument can be specified via the
num_acceptor transport option. Ranch has been updated
to 1.4.0 to that effect.
This option allows customizing the compacting of the Req object
when using Websocket. By default it will keep most public fields
excluding headers of course, since those can be large.
This callback is called when an error occurs before the request
(including headers, excluding body) was fully received. The
init/3 callback will not be called. The callback receives the
partial Req object (possibly empty), the reason for the error
and the response command that the server will send. It allows
you to be aware of the error and possibly modify the response
before it is sent.
These tests cover frame sizes. It's mostly edge cases for sure
(ie misbehaving clients and us having to reject them properly).
I had these almost ready for a long time, so I'm glad I can
push them out.
This requires updating Cowlib too (we currently track master).
