Now previous padding options are removed and instead we added
select_padding_callback to select padding length for each frame
by application. If this callback is not implemented by application,
no padding is added.
This change also fixes the broken session_detect_idle_stream()
if stream_id is our side.
This stream inflater can inflate incoming header block in streaming
fashion. Currently, we buffer up single name/value pair, but we chose
far more smaller buffer size than HTTP/2 frame size.
nghttp2 library itself now accept octet header/value pairs,
completely not restricted by HTTP/1 header name/value rule.
The applications may impose restriction about them using
validators.
The NGHTTP2_ERR_PAUSE library error code is introduced to pause
the execution of nghttp2_session_mem_recv() when that error code
is returned from nghttp2_on_frame_recv_callback or
nghttp2_on_data_chunk_recv_callback. If this happens, the parameters
available for both callbacks are retained until the application
calls nghttp2_session_continue(). The application must retain
input bytes which was used to produce the frame.
After successful call of nghttp2_session_continue, the application
can continue to call nghttp2_session_mem_recv() to process
additional data.
This function is similar to nghttp2_submit_request and the
difference is it takes an array of nghttp2_nv as name/value pairs.
It is useful if name/value pairs is not NULL-terminated in the
application code.
Fix the bug that connection-level local window is not updated
for the data is the last part of the stream. For the stream
level window may ignore this, connection-level window must
be updated. Also this change fixes the bug that connection-level
window is not updated for the ignored DATA frames.
SPDY spec does not allow multiple empty header values in one header
field. This change makes out-going framer ignore such empty header
value if there is non-empty header value with the same name.
This change implements spdylay_map functions in iterative approach
instead of original recursive approach. The iterative version is
overall faster than recursive one. The traversal function
(spdylay_map_each) is slightly slower but it is not noticeable.
If resultant length of a frame exceeds the maximum value (which is
2**24 - 1 for SPDY/2 and 3), SPDYLAY_ERR_FRAME_TOO_LARGE is used to
indicate this error. This error will be notified by
on_ctrl_not_send_callback.
spdylay_npn_get_proto_list() returns a pointer to the supported SPDY
version list. The element of the list is spdylay_npn_proto struct. It
contains all SPDY version information this library supports. The
application can use this information to configure NPN protocol
offerings/selection.
We say the header block is invalid if at least one of the following
condition is true:
There are duplicate header names; or the header names are not
encoded in US-ASCII character set and not lower cased; or the
header name is zero-length string; or the header value contains
multiple in-sequence NUL bytes.
spdylay_frame_unpack_nv() returns SPDYLAY_ERR_INVALID_HEADER_BLOCK
if the unpacking suceeded but it found the header block is invalid.
This means that caller treats it as success, but do additional
processing for invalid header block if it wants.
The functions calling spdylay_frame_unpack_nv() also return
SPDYLAY_ERR_INVALID_HEADER_BLOCK.
Now remote and local settings are stored separately.
The initial window size is included in SETTINGS frame, all active stream's
window sizes are now updated.
Removed the initial_window_size member from spdylay_stream because
it is the same as remote_settings's initial window size.
This callback function is invoked after the control frame
is not sent because of the error. The error is indicated by
the error argument, which is one of the values defined in spdylay_error.
This function invoked when request from remote peer is
received. In other words, frame with FIN flag set is received. In
HTTP, this means HTTP request, including request body, is fully
received.
If max concurrent streams limit is reached, SYN_STREAM frames are not sent
and backed off. If other type of frame is waiting in the tx queue, it is
sent first. We introduced another priority queue for this purpose.
In this change we did not add code to send RST_STREAM when SYN_STREAM is
received but max concurrent stream is reached.
We use just single buffer to store name/value headers fields, instead of
allocating memory for each name/value strings.
It is now more than 2 times faster than old one.