TNonblockingServer and TFramedTransport used to have the option to only
frame messages on one side of the communication. This capability was
removed from TFramedTransport because it was poorly implemented and not
useful. This change removes it from TNonblockingServer as well, and
removes references to it in some of the C++ test code.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@673791 13f79535-47bb-0310-9956-ffa450edef68
This patch adds two Thrift servers for Java that both use non-blocking I/O
to avoid locking up worker threads for idle connections. The two classes are
- TNonblockingServer, which supports single-threaded serving
- THsHaServer, which performs I/O in one thread and method invocations in
a configurable thread pool.
To support these servers, TNonblockingServerSocket and TNonblockingSocket
have been added.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@673550 13f79535-47bb-0310-9956-ffa450edef68
- Parse integer constants in Thrift files as 64-bit ints.
- Die if an overflow occurs.
- Warn if an enum value cannot fit in 32 bits.
- Add a simple test for the above.
I ran all of the generators over BrokenConstants.thrift before adding the
overflow, and they appeared to work. However, the code generated was not
always valid (for example, the 64-bit constant must have an LL suffix in C++).
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@672907 13f79535-47bb-0310-9956-ffa450edef68
I changed the behavior of TFramedTransport when flush is called without
writing any data to the transport. This broke the unit test, which was
relying on a weird corner of TFramedTransport's behavior in order to do
some stricter checking. I altered the unit test to never flush without
writing and added a new test to verfy the "empty flush" behavior.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@672881 13f79535-47bb-0310-9956-ffa450edef68
- Clean up lib/cpp/Makefile.am to use AM_CXXFLAGS etc instead of
target-specific so we can take advantage of default flags.
- Use top_builddir in a few places instead of top_srcdir.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666489 13f79535-47bb-0310-9956-ffa450edef68
- Modify configure.ac to check for Python conditionally, like we do for
Java, Mono, and Erlang.
- Modify Makefile.am to add "py" to SUBDIRS only if Python was detected.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666488 13f79535-47bb-0310-9956-ffa450edef68
Summary:
This is to make it easy to run Hadoop mapreduces using Hadoop Streaming on thrift-serialized structs
without implementing any special file splitter or anything
Test plan: test_disklog:t_base64()
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666466 13f79535-47bb-0310-9956-ffa450edef68
Summary:
This is for logging applications with thrift_disk_log_transport, so the function calls
logged don't necessarily have to be async void
Test plan:
Added to test_disklog.erl
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666465 13f79535-47bb-0310-9956-ffa450edef68
Summary:
Throws a 'TApplicationException' record
Test plan:
Changed cpp TestServer so that if testException is called with "ApplicationException" as an argument, it throws a TException which is serialized back as an EXCEPTION type message
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666421 13f79535-47bb-0310-9956-ffa450edef68
Summary:
- reply_type for async void functions is now async_void instead of
the empty struct definition
- async void functions should return "ok". otherwise the processor
will crash and the connection will be killed. Is this behaviour
expected?
Test Plan: tested using testAsync() in ThriftTest. Didn't used to work
but works now
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666415 13f79535-47bb-0310-9956-ffa450edef68
When a function called by the forking python thrift server throws an
exception the client will hang. This happens because the parent of the
forked process does not properly close the socket fd.
Under normal operation the server operation completes and returns a value to
the client. However, when an exception occurs the 'end' message is never
sent to the client so the client relies on a connection close to abort the
call (this is how the threading server works I believe).
This fails with the forking server because the parent process never closes
the socket fd. The child has closed the fd at this point, but the
connection will not actually be closed until all open instances of the fd
are closed. Since the client is waiting for a message and the server never
closes it the client is forced to wait until a read timeout occurs many
minutes later.
This diff closes the parent's copy of the socket fd immediately after the
fork occurs.
I modified my load test server to throw an exception. The client did not hang.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@666363 13f79535-47bb-0310-9956-ffa450edef68
The Java Makefile.ams assume the existence of ANT, but "make distclean"
runs in the Java directories even if ENABLE_JAVA is off. This change
makes the clean-local target handle the absence of ANT, fixing
"make distclean".
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665683 13f79535-47bb-0310-9956-ffa450edef68
The old implementations of the memory-based transports (TBufferedTransport,
TFramedTransport, and TMemoryBuffer) shared very little code and all worked
a bit differently. This change unifies them all as subclasses of a single
base (TBufferBase) which handles the fast-path operations (when requests
can be satisfied by the buffer) with inline methods (that will eventually
be made nonvirtual in the template branch) and calls out to pure-virutal
methods to handle full/empty buffers. All of the buffer-management is now
done in terms of "base and bound" pointers rather than "pos" integers.
These classes were moved to TBufferTransports.{h,cpp}. The .h is included
in TTransportUtils for backwards compatibility.
Also added a "TShortReadTransport" to assist testing transports.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665676 13f79535-47bb-0310-9956-ffa450edef68
Also throw an error in the compiler if we cannot generate a constant for a
declared const because of its type. Added a test of this functionality in
ConstantsDemo.thrift.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665675 13f79535-47bb-0310-9956-ffa450edef68
Summary: I neglected to make the call to readMessageEnd if the client read an app exception in the haskell and ocaml code. Fixed now. The test cases for both now also compile with the new ThriftTest.thrift (though haskell's is still not doing anything but echoing the args)
Reviewed By: dreiss
Test Plan: Ran test code. This bug would actually never show up because there's only tbinaryprotocol implemented for these languages and that does nothing on a readMessageEnd.
Revert: OK
DiffCamp Revision: 11377
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665651 13f79535-47bb-0310-9956-ffa450edef68
Previously, TPipedTransport wrote it's full buffer to the "pipe" and
fully reset its buffer on a readEnd. This assumed that the buffer was
fully read at that point. This is not the case if requests are pipelined.
This change makes it only pipe out the portion that has been read and
copy the unread portion to the beginning of the buffer.
Also add a test that verifies the new functionality.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665646 13f79535-47bb-0310-9956-ffa450edef68
- Make the Perl generator use program->get_namespace("perl")
instead of program->get_perl_package()
- Eliminate the explicit "perl_package" in t_program.
- Deprecate the perl_package token.
- Update example .thrift files and syntax files.
git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@665624 13f79535-47bb-0310-9956-ffa450edef68
