notion of a global group (though we implicitly define one in InvocationCodes)
added a mechanism for directors (which generally handle the client side of
invocation services) to register their interest in bootstrap service groups so
that the whole goddamned complex business can happen magically behind the
scenes.
If you instantiate a director, it will automatically register interest in the
service group it needs and everything will work. If you don't use the director
code, you don't get the services and you can safely exclude all of that code
from your client even though the services are still in use on the server (and
presumably used by some other types of clients).
This is going to break all the builds, which I'll soon fix. Then I'll go write
all this in ActionScript. Yay!
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4552 542714f4-19e9-0310-aa3c-eee0fc999fb1
MetaSOY clients (Swiftly, World, and soon Admin Dashboard) don't have to know
about a bunch of unrelated crap. Fricking complexity++, grumble.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4551 542714f4-19e9-0310-aa3c-eee0fc999fb1
subinterfaces of InvocationProvider and forward requests to peer nodes
in order to use the same methods on owning peers and cloning peers.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4546 542714f4-19e9-0310-aa3c-eee0fc999fb1
queue up an invoker unit which would go off and do some database stuff and then
by the time it came back and was ready to publish its results to a distributed
object, the object in question would have been destroyed for any of a variety
of fairly natural reasons (client disconnected or logged off, game was
abandoned, dog ate homework).
One "solution" to this problem would be to litter our games' code with
thousands of calls to isActive() in the handleResult() methods of our invoker
units. We've done a bit of that in Yohoho but I've resisted starting down that
path in our other games.
Another solution would be to create an Invoker.Unit wrapper that takes a
reference to the distributed object (or objects) that it will be modifying and
have the common unit code check that the object(s) in question are still alive
at the end of the asynchronous operation and not call handleResult() if they
are not. This has numerous problems: what do you do if one object is alive but
not another, how do you incorporate this functionality in with the numerous
other Invoker.Unit derivations we have that simplify our lives in other ways
(without getting crazy and starting to use something like AOP), do you silently
abort the operation or log something?
So instead, I've come around to the idea that this is simply a dirty fact of
life in asynchronous programming and the fact that we can accept modifications
to distributed state after the distribted object in question is dead is a good
thing. We used to log a warning every time this happened and freak out even
more substantially if one tried to start a transaction on a dead object. Now we
will simply log an informational message (I don't think this sort of thing
should be silently ignored because there are some cases where it is an
indication of incorrect code, those are simply more rare). We will also allow a
transaction to be started on a dead object and when the transaction is
committed, all the events involved will be dropped just like a single
modification would have been dropped on that object.
This allows the most sensible thing to happen which is any results that are
published to still live objects will actually be published and results
published to dead objects will be dropped without making a big fuss. Since a
dead object by definition cannot have subscribers, no one could possibly have
cared about the dropped events anyway.
Also widened.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4545 542714f4-19e9-0310-aa3c-eee0fc999fb1
acquire and release locks, and to use handler objects to manage locks in
the process of resolution.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4542 542714f4-19e9-0310-aa3c-eee0fc999fb1
exposes the need to write a Client that is optimized to act as a proxy instead
of just using the one that is designed to run on a user's machine. However,
this will be fine for now. Integration with PeerManager forthcoming.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4530 542714f4-19e9-0310-aa3c-eee0fc999fb1
response constants start with E_ and their codes start with e..
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4491 542714f4-19e9-0310-aa3c-eee0fc999fb1
(ByteArray is a special class for dealing with bytes. It does not extend
the normal Array).
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4452 542714f4-19e9-0310-aa3c-eee0fc999fb1
the client with the proper source oid. Normally this happens when the client's
event is received over the network but in local mode events never go over the
network and thus all appear to have originated on the "server".
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4435 542714f4-19e9-0310-aa3c-eee0fc999fb1
an arbitrary Java class which we are converting to ActionScript (the language,
not the class). Hopefully this will appease the compiler.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4419 542714f4-19e9-0310-aa3c-eee0fc999fb1
aren't magically in scope for ActionScript, we have to generate import
statements for them.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4414 542714f4-19e9-0310-aa3c-eee0fc999fb1
when we encounter the pattern "new SomeClass[0]". Le whee!
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4410 542714f4-19e9-0310-aa3c-eee0fc999fb1
field/method names for the ActionScript version of a class (or to omit
something entirely). This removes the need for special case hackery for
toStringBuilder().
In order for annotations to work, however, we have to require that the
GenActionScriptTask be loaded from the same classloader that loads the classes
to be reflected upon. Before we only reflected on the target classes, never
instantiated them. Annotations are actually instantiated, so we have to be able
to create an instance of the ActionScript.class that is compiled into our
target classes and assign it to a reference that is compiled into
GenActionScriptTask. Beware the complexities of dealing with multiple class
loaders.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4405 542714f4-19e9-0310-aa3c-eee0fc999fb1
because we can't do that in ActionScript; properly handle static var and
non-static const.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4404 542714f4-19e9-0310-aa3c-eee0fc999fb1
interfaces, and Marshallers for same. The remaining snag has to do with the
annoyance of ActionScript not supporting inner classes, which means that
ChatService.TellListener for example has to become ChatService_TellListener.
The code for generating the Java marshaller knows to add an import for
ChatService if some random invocation service interface happens to reference
ChatService.TellListener, but now it needs to be made to know to add an import
for ChatService_TellListener in ActionScript land and it has to do it in a way
that doesn't fuck up the Java code generation. Whee!
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4397 542714f4-19e9-0310-aa3c-eee0fc999fb1
Java. Of course this is a very limited translation facility that is mainly
focused on auto-generating Streamable ActionScript classes from their Java
originals, and it doesn't actually convert method bodies, just field and method
declarations, initial values, and such.
Most of our Streamable classes don't have much in the way of real methods, and
the autogenerator will stick new methods in when they show up on the Java side
so at least we'll *know* that they need to be implemented. And the main thing:
readObject and writeObject are in fact implemented by the code generator so at
least we won't be plagued by annoying streaming errors when we change something
on the Java side and forget to change the ActionScript side.
Next up, auto-generating InvocationService interfaces and InvocationMarshaller
implementations in ActionScript and ActionScript versions of DObjects.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4394 542714f4-19e9-0310-aa3c-eee0fc999fb1
checkin, tihs will be a non-issue when we start devoting an entire separate JVM
to each game (which we will have to do for all games that run server code)
because then all a game can do is stick a fork in its own JVM and render itself
unplayable.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4388 542714f4-19e9-0310-aa3c-eee0fc999fb1
stick a fork in the server otherwise malicious or broken code can generate an
AbstractMethodError or some other error and cause the server to decide to quit.
It may still be possible for malicious code to throw an OutOfMemory error, and
if that's the case then I don't know what we can do to both gracefully get the
fuck out of dodge when things are pear shaped and avoid the wrath of the
kiddies.
Incidentally, it was stupidity, not malice that brought this to my awareness.
git-svn-id: svn+ssh://src.earth.threerings.net/narya/trunk@4387 542714f4-19e9-0310-aa3c-eee0fc999fb1
Michael Bayne