Changed HTTPServerConnection to return a Runnable to be run by the
execution strategy also in case of content.
This allows onDataAvailable() to be called at the proper times.
Ported the "maxAssociations" functionality from SPDY's ReferrerPushStrategy.
Added JMX support.
Removed __renew__ special path in favour of a JMX method.
Added clearPushCache() JMX method.
Made push reentrant by eliminating the check for "org.eclipse.jetty.pushed".
Before, the sender was updating the window size after the SETTINGS
frame was written.
This was leading to a race where the receiver saw the updated window
size and sent DATA frames; these were received by the original sender
before it had the chance to update its local window size, causing an
error.
Now, the update of the window size happen just before writing the
SETTINGS frame to avoid this race.
Fixed HttpChannel to avoid to stack overflow in case of unusual
exceptions thrown while committing.
Now it directly calls the transport to try to send a 500.
Testing for stream.isClosed() is done in StreamCloseTest.
The tests were failing because receiving the headers does not mean
the sending the headers has notified the callback yet, and hence
closed the stream.
Moved the sneaky direct upgrade to its own method and added support to detect a standard upgrade.
The standard upgrade now creates the h2c connection and decodes the settings header, but it does not yet process the
settings nor send a response to the request.
Fixed by tracking both send and recv initial stream windows.
This is needed because both client and server may send an
INITIAL_WINDOW_SIZE setting, and they must be treated
separately.
Naked writes could throw WritePendingException.
Now we wait until the client has finished sending the reply to the
server SETTINGS frame, then we do the naked write.
The client connection's promise was succeeded in the context of a
call to SelectorManager.newConnection().
This was wrong, since succeeding the promise may trigger the send
of a request *before* the connection is actually linked to the
endPoint and opened.
This change moves the succeeding of the client connection's promise
to the connection onOpen() method.
+ HttpField.nameHashCode() fixed to actually be US-ASCII case
insensitive per documentation
+ Since removal of MetaData equals/hashcode, the comparison
of the MetaData and MetaData.Response is now done entirely
within the HpackTest
Now the stream close state is updated when the frame has been
successfully written, and when it is received.
The stream is closed in case of failures.
Just after the stream close state update, if the stream is closed
then it is removed from the session.
It is not possible to perform asynchronous processing of the content
of DATA frames, because otherwise the parser has to stop, stalling
all other streams.
Parser.Listener methods were returning boolean in a vestigial attempt
to handle asynchronous data processing, and have now been converted to
return void.
fillInterested() is called.
This is needed to avoid race conditions where fillInterested()
triggers a new thread entering onFillable() and acquiring a new buffer
while the previous thread is releasing the previous buffer.
Now instead of having the channel to dispatch when it detects that it
has to call the application (upon receiving a HEADERS frame, or upon
a push "fake" request), now the whole mechanism is controlled by an
ExecutionStrategy.
This is an initial recyling of the channel. The pool is a concurrent linked list, but it may be better to use an array backed list/stack that can easily enforce a maximum pool size.
The pool is currently on the session listener, when it really should be on the HTTP2ServerConnection, but more refactoring is needed for that.
Also the pool is currently inaccessible to the push mechanism
RequestLog can now be set on the HttpChannel and is called during the onCommitted callback.
The RequestLog may be set either on the server or via an updated RequestLogHandler that can
be applied to a specific context.
The HttpContent class has been reworked to store HttpField instances, we may be generated on
demand or instances of PreEncodedHttpField.
The encoding of HTTP2 fields has been generalized to handle both indexed and literal fields, selected
by header enum set.
Default servlet and response classes have been cleaned up in how they set response headers.
Introduced parameter "dispatchIO" in the relevant factories so that
they can be configured by users and connections will be created
taking into account this parameter.
For less configurable connection factories, this parameter is
currently hardcoded to either true or false depending on the case.
For example, ALPN and NPN connections have it to false, since they
don't do any blocking operation in onFillable().
Fixed by ensuring that when a failure happens, either by catching an
exception or by failing a callback, we always call completed() and
abort the channel (via new method terminate()).
Introduced ResetException, and using it when failing frames of streams
that have been reset already.
HttpTransportOverHTTP2.abort(Throwable) checks for this exception and
does not close the connection.
A PushCacheFilter contains the logic to associate secondary resources
to primary resources.
PushCacheFilter calls a Jetty-specific API on the request dispatcher:
Dispatcher.push(ServletRequest). This is a technology preview of the
push functionality slated for Servlet 4.0.
The push() invocation arrives to the transport and it is converted to
HTTP/2 specific PUSH_PROMISE, along with the mechanism to simulate
the request for the secondary resource.
No more need to copy protonego-alpn.xml to a jetty.base to specify
which protocols are advertised and in which order.
Instead, simply specify:
alpn.protocols=h2-14,http/1.1
alpn.defaultProtocol=http/1.1
in start.ini (in the example above for http2).
No more need to copy protonego-alpn.xml to a jetty.base to specify
which protocols are advertised.
Instead, simply specify:
protonego.protocols=h2-14,http/1.1
protonego.defaultProtocol=http/1.1
in start.ini (in the example above for http2).
It's now sent after a call to onPreface(), which has been moved to
the common interface Session.Listener (from ServerSession.Listener),
so that client applications can customize the SETTINGS to send to the
server.
Flow control window updates are now processed by the flusher, so that
it is the only component that handles window updates.
In the process of this refactoring, HTTP2Flusher was refactored out
of HTTP2Session.
There is a difference between the value set via configuration, that
always refer to remote streams (streams initiated by remote peers),
and the value received via SETTINGS frame, that always refer to local
streams (streams initiated locally).
The case was that shutdown was called, ShutdownFlusherEntry called
flusher.close(), which called super.close(), which called
onCompleteFailure(), which looped over the active items to fail them,
calling again ShutdownFlusherEntry, which called again flusher.close(),
etc.
This refactor strives to remove duplication between the Metadata class and the HttpURI class.
Both classes have been made mutable (as they partially were anyway so best not to pretend).
HttpURI now holds the decomposed strings rather than a single string with indexes. This allows it to be rebuilt after changing just parts of the URI. It is now a lot more similar to the
JVM URI class and we could consider replacing it (after checking peformance).
Next step is to refactor the Request class to prevent it duplicating these fields.
Simone Bordet
Joakim Erdfelt
Greg Wilkins
Jan Bartel
Jesse McConnell