AfflictionNetworks/narya
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Commit Graph

  • dd43d89028 A few extra lines of code to support a fourth enum policy: name only. Ray Greenwell 2011-03-08 18:30:50 +00:00
  • b4b409caa3 Provide an option to show all command aliases in the usage text and only show the first alias in the big list. Andrzej Kapolka 2011-03-08 04:57:43 +00:00
  • 08f6dbefa9 Comments. Ray Greenwell 2011-03-08 01:08:33 +00:00
  • 55f031304c Let's rename those enum policy constants based on how they work, not whether they're old or new. OLD -> NAME_WITH_BYTE_ENUM NEW -> ORDINAL NEW_WITH_BYTE_ENUM -> ORDINAL_WITH_BYTE_ENUM Ray Greenwell 2011-03-08 01:01:06 +00:00
  • b1d7577599 - Cleaned up streamers. - Allow configuring of two new streaming policies for enums. - Changed the alphabetical fields policy to be checked once on class init. Ray Greenwell 2011-03-08 00:48:39 +00:00
  • 3b5e4d3f68 Fix the formatting of that comment... Mike Thomas 2011-03-07 21:16:52 +00:00
  • 24d126dd63 Allow the streamer to be used in a jvm-neutral way, streaming fields in alphabetical order. Sun JVMs use the order they're defined in the file, IBM JVMs apparently use the reverse of that, and Dalvik/Android JVMs stream in alphabetical order. Ideally, the alphabetical mode would be the default but this would break all kinds of legacy stuff, so I'm not going to make that the default. (I had done this before on the narya-1.5 tags - doing correctly on trunk now.) Mike Thomas 2011-03-07 21:07:25 +00:00
  • 3640b7c289 The INSTANCE constant only really makes sense for ListStreamer, as it's used for Collections, too. Ray Greenwell 2011-03-07 19:58:48 +00:00
  • 568765dd13 Whitespace Dave Hoover 2011-03-07 19:55:59 +00:00
  • 83d24a6d6e Update the comment, too Dave Hoover 2011-03-07 19:55:58 +00:00
  • a46e794730 Iterable is now supported, even more basic than Collection. Ray Greenwell 2011-03-07 19:52:41 +00:00
  • b0b4b5912c Get a little crazy with the Cheez Whiz and use our shared build system for narya. Charlie Groves 2011-03-06 23:11:08 +00:00
  • 490b381127 Streaming for Multiset and Iterable. Ray Greenwell 2011-03-05 15:31:46 +00:00
  • 53235c5212 Preparing for 1.8-SNAPSHOT development. Jamie Doornbos 2011-02-28 20:01:14 +00:00
  • 5a65b218d8 Preparing for 1.7 release. Jamie Doornbos 2011-02-28 19:53:19 +00:00
  • f3c0359953 Allow users of the PersistingUnit to set the desired result from the invoke method rather than having to also override handleSuccess. Jamie Doornbos 2011-02-28 19:30:34 +00:00
  • 81bfee5132 Generate receivers and decoders for actionscript Charlie Groves 2011-02-26 02:04:06 +00:00
  • c3d9b3b238 Move these variables where they belong and add some documentation. Also apparently whitespace. Tom Conkling 2011-02-21 23:05:08 +00:00
  • e7a725f9e1 Preparing for 1.7-SNAPSHOT development. Jamie Doornbos 2011-02-21 22:02:35 +00:00
  • 202e3fa49d Preparing for 1.6 release. Jamie Doornbos 2011-02-21 21:55:41 +00:00
  • 061b55815d Just expose this number in case anyone needs it Jamie Doornbos 2011-02-21 21:36:19 +00:00
  • 57b0175f7d Allow chat history to be held just long enough for the current stack to unwind. Jamie Doornbos 2011-02-21 19:23:24 +00:00
  • cbb97250ac Missed a spot Jamie Doornbos 2011-02-21 18:35:08 +00:00
  • ee81934087 Extract the chat history code into a separate injectable module. This will make it easier to add new features or change the implementation. Jamie Doornbos 2011-02-21 18:34:06 +00:00
  • b5bd9ece3c Preparing for 1.6-SNAPSHOT development. Mike Thomas 2011-02-17 19:43:10 +00:00
  • 01e7f8ef04 Preparing for 1.5 release. Mike Thomas 2011-02-17 19:38:55 +00:00
  • eb41f8b70b - The secret would not be properly assigned to the ClientLocal during session resuming (and likely on peer transfers of client connections), so have PresentsSession handle that in both starting and resuming a session (and takes out a synchronized call in ClientManager). Mark Johnson 2011-02-15 18:42:46 +00:00
  • ee2c653c3e - This secret ended up being stored in the auth request Mark Johnson 2011-02-15 18:24:19 +00:00
  • 7a77e9c9d0 NPE? FML. Andrzej Kapolka 2011-02-15 05:13:30 +00:00
  • 4a5fe35c2c - Save the aes key generated during secure authentication in Client and ClientLocal (for the server). It can then be used to encrypt further data during the session. Mark Johnson 2011-02-15 01:20:34 +00:00
  • acc6cff7e5 'i' before 'e' except in comments Matt Jensen 2011-02-12 01:11:21 +00:00
  • 360dee3de8 Missed a spot Charlie Groves 2011-02-10 23:41:14 +00:00
  • d40dad55fe Unused imports Charlie Groves 2011-02-10 23:40:20 +00:00
  • 0a5751239f Whoops, fix import ordering Tom Conkling 2011-02-09 20:52:05 +00:00
  • 7dc50f23d6 Make Streamer.registerStreamer public, and add a warning if an existing Streamer is replaced. Tom Conkling 2011-02-09 20:50:52 +00:00
  • f18cd38c3d Try again - instead of a major reorganization, simply extend the range of conversions. An [B is always a ByteArray, and after that, any array becomes a TypedArray. Par Winzell 2011-02-08 23:26:32 +00:00
  • 25f1eac53d Reluctantly back out 6484. It wasn't quite that simple. The problem appears to be that the AS3 the templates actually end up generating depends on the original sussed-out Java types in sometimes non-obvious ways; for example, parameter types collapse bytes, characters, shorts and ints to a mere 'int', without us really being able know in advance, so we will be generating superfluous e.g. util.Byte imports. A more elegant solution might be to construct Java imports and AS imports side by side in the earlier phases. Par Winzell 2011-02-08 22:57:00 +00:00
  • 9a985c0629 Assemble all the ad-hoc translations of primitive (and other) Java types to AS3 variants in one method and call that method. This should hopefully end us up with generated service listeners that compile. Par Winzell 2011-02-08 22:02:17 +00:00
  • f08f8158f0 Pass the mode into mogrifyChat. Andrzej Kapolka 2011-02-08 01:55:23 +00:00
  • 8a1ff4ac4b RFC: providing a way to generically detect the language and machine identity for a connection Jamie Doornbos 2011-02-07 23:15:15 +00:00
  • 5df4dfe91c Handle Comparable too. Par Winzell 2011-02-07 21:01:32 +00:00
  • d00d0ca5da You can now specify a client must require a secure channel to authenticate. If this is set, {Client.setRequireSecureAuth} then if the client fails to create a secure channel with the server, an auth request with no credentials will be sent. The supplied Authenticator will need to handle this situation to send the appropriate response back to the client (ie: tell it to download the latest version). Mark Johnson 2011-02-04 23:14:45 +00:00
  • 92c5163707 - Have the PublicKeyCredentials provide the secure version being used on the client, so when we eventually change how the handshake works, we can maintain backwards compatability with the old clients Mark Johnson 2011-02-04 22:10:16 +00:00
  • e93459bdb9 Fix some 1.6isms and add a missing @Override Mike Thomas 2011-02-04 21:33:39 +00:00
  • ae6c134546 An optional secure channel can be created for authentication to ensure credentials are encrypted when sent to the server. Mark Johnson 2011-02-04 19:38:53 +00:00
  • 540f84194d Break out a method to find all applicable nodes for a NodeRequest, and add a class to hold a node name and an oid to fully address a DObject in a peered environment. Charlie Groves 2011-02-04 04:46:17 +00:00
  • 5b6c9a71e5 Fix a bit of a memory leak wherein if a DObject happened to be destroyed prior to its clients unsubscribing, that DObject would be kept until the sessions still subscribed to that object logged off. Due to various client-side timing bits this could happen even if the server sends a message to clients that causes them to unsubscribe prior to calling destroyObject. Mike Thomas 2011-02-03 19:28:34 +00:00
  • 04df68ea68 Add an IntervalFactory interface, which just declares newInterval(), and have RootDObjectManager extend it. Tom Conkling 2011-02-02 01:27:38 +00:00
  • 084f5ba944 Fixed typo. Michael Bayne 2011-02-01 19:37:59 +00:00
  • c19b395a43 Name normalization has been space sensitive for nearly 7 years now. Stop saying otherwise. Charlie Groves 2011-01-31 07:08:12 +00:00
  • 8b3bb03751 Import fix. Par Winzell 2011-01-30 05:43:03 +00:00
  • 2c17d66f02 Add the messageReceived and destroyed signals to DObject rather than generating them. If they were generated, subclassing would fail on field conflicts. Charlie Groves 2011-01-28 19:33:46 +00:00
  • a381c8544e We only need ooo-external for these now Charlie Groves 2011-01-28 19:32:30 +00:00
  • 45879635a3 Let's preserve binary compatibility with extant DObjects for a little while. Michael Bayne 2011-01-27 20:12:40 +00:00
  • 871cc2eb63 Avoid a ClassCastException when we transfer a marshaller from client to server (as we sometimes do when we transfer a player between peers in Project X). Technically we probably shouldn't be streaming the marshallers in that case, but doing this (and clearing the marshaller on the receiving end) is simpler than clearing, streaming, and restoring; or settings things up so that the marshaller is specifically excluded from streaming in transfers. Andrzej Kapolka 2011-01-24 23:41:29 +00:00
  • fb822af0e9 It's probably worth including the stack trace for this warning, since it can come from various different places. Andrzej Kapolka 2011-01-24 22:19:40 +00:00
  • 6d36a54999 I don't think this will break anything: provide a means for names to override the default comparison methods in a symmetric way (so that it doesn't matter which of the two objects' equals/compareTo methods was called). Andrzej Kapolka 2011-01-20 05:06:54 +00:00
  • 92c1225d0e Deploy a sources jar along with the code jar for IDE-happiness Charlie Groves 2011-01-20 00:40:12 +00:00
  • 1463fe0fce Step 2 in getting the *-poms versions to match pom versions. Mike Thomas 2011-01-18 17:55:31 +00:00
  • 4bbc8adaf1 And move the tools-pom in line too. Mike Thomas 2011-01-18 17:47:13 +00:00
  • 193506a7c7 Step 1 in bringing gwt-pom up to pom's version. Mike Thomas 2011-01-18 17:41:39 +00:00
  • 89b7a4df80 Step 1 in bringing as-pom up to pom's version. Mike Thomas 2011-01-18 17:37:02 +00:00
  • 4c44d7063b Increment version to 1.5-SNAPSHOT Mike Thomas 2011-01-14 19:43:57 +00:00
  • 266cb13db1 Set this as narya 1.4 Mike Thomas 2011-01-14 19:31:05 +00:00
  • 52afa1d654 Use m2eclipse for Eclipse configuration Charlie Groves 2011-01-11 23:46:41 +00:00
  • 50b19cdee0 Unnecessary import Charlie Groves 2011-01-11 18:53:59 +00:00
  • 5c3e0ed825 Make OidList Iterable Tom Conkling 2011-01-10 22:47:09 +00:00
  • 0e8f5e4da4 If we generate arg1, ... parameters but arg0, ... arguments, things go badly. Par Winzell 2011-01-10 20:27:30 +00:00
  • 5b2ee423bf This is why typechecking is a good thing. Par Winzell 2011-01-07 20:58:42 +00:00
  • d66ed6f7fe This method is in fact defined on the listener. Par Winzell 2011-01-07 20:58:17 +00:00
  • aff817a70f Charlie notes that if the old code worked, we can surely eliminate the setManager call. Ray Greenwell 2011-01-06 23:51:56 +00:00
  • bc5303b0fa This must not have been a bug, but it sure looks like one. Ray Greenwell 2011-01-06 23:42:35 +00:00
  • 8f8a50c9e8 Apply additions and removals to OidList fields on DObjects immediately on the server to make it match DSets and setting attributes. With the old behvaior, if a client subscribed to a DObject and modified an OidList on it in a single pass, it could miss the modification. The DObject is serialized immediately when the server gets the subscription request, but events aren't sent for that subscription till it's processed the next time the omgr queue comes round. Charlie Groves 2011-01-06 09:15:47 +00:00
  • b4af8a0384 Include the traceback when we fail to extract a name from source Charlie Groves 2011-01-06 09:09:01 +00:00
  • edf6dccdf7 Generate OidList signals for actionscript Charlie Groves 2011-01-06 09:08:24 +00:00
  • 2572112560 Need to splice the arguments, too. Andrzej Kapolka 2011-01-06 03:30:51 +00:00
  • 5a66c46308 Yeah, thanks for that. Andrzej Kapolka 2011-01-06 03:25:09 +00:00
  • 0dae31b2c9 ActionScript DObjects dispatch a signal when they receive a MessageEvent Tom Conkling 2011-01-05 23:49:45 +00:00
  • f9ce9f6939 On second thought, let's configure the standalone client reference via an optional injection. Andrzej Kapolka 2011-01-05 02:03:25 +00:00
  • 0c28d31357 Initializing the invocation director reference on deserialization doesn't work well in standalone mode. We need a way to install a director on registration. Andrzej Kapolka 2011-01-05 01:32:13 +00:00
  • 68aba0e92a Add a mapping from java.lang.Cloneable -> com.threerings.util.Cloneable Tom Conkling 2011-01-05 00:21:16 +00:00
  • 49c6511fd1 Because getInterfaces only returns the interfaces implemented by the class (not by its superclasses), we need to rise through its superclasses as well in order to find the provider interface. Andrzej Kapolka 2011-01-04 22:46:22 +00:00
  • 6cba470736 These override their superclass, so annotate as such. Mike Thomas 2011-01-04 19:35:43 +00:00
  • bfd1f79107 Bugfix: we need to examine methods from superclasses as well. ProjectXZoneRegistry extends a class that is a ZoneProvider... Ray Greenwell 2011-01-03 21:17:55 +00:00
  • 55d1df93a1 Avoid ArrayIndexOutOfBoundsException. Andrzej Kapolka 2011-01-01 21:52:23 +00:00
  • 02899d1083 Auld Lang Syne Dave Hoover 2011-01-01 05:02:21 +00:00
  • f43c511436 No longer need to generate dispatchers. I'll leave the code here for the time being. Michael Bayne 2011-01-01 01:33:07 +00:00
  • 97be2d9fc7 These guys are no longer needed. Michael Bayne 2011-01-01 01:32:34 +00:00
  • f8c7241f9f Switched providers to new dynamic dispatching. Michael Bayne 2011-01-01 01:31:20 +00:00
  • 5a67aa906b A while ago, I put on my todo list: "look into changing InvocationService to use JVM generated proxies, nix marshaller/dispatcher." Michael Bayne 2011-01-01 01:29:26 +00:00
  • 1db66920d2 Deprecated DSet.toArray. Pretending like arrays are parameterized collections is a bad idea and will lead to pain. Nixed the one place where toArray was used (which was a serous pig's breakfast itself). Michael Bayne 2011-01-01 01:15:33 +00:00
  • 328ff30506 Marshaller no longer needs Client import. Michael Bayne 2010-12-31 22:42:45 +00:00
  • 127e388349 Unused import nixage. Michael Bayne 2010-12-31 22:36:46 +00:00
  • 90c6429070 Missed a spot. Michael Bayne 2010-12-31 22:32:15 +00:00
  • e80409a7f4 Updated to use new Client-reference-free style. Michael Bayne 2010-12-31 22:28:21 +00:00
  • b60112ae93 Some fiddling to keep the CPP bits in sync as well. Michael Bayne 2010-12-31 22:27:51 +00:00
  • 5d2852544b Generate our test service(s) as well. Michael Bayne 2010-12-31 22:17:00 +00:00
  • 36a98f8794 Removed the need to supply a Client reference when making invocation service calls on the Java side, to match this added convenience we introduced to the ActionScript side some time ago. Michael Bayne 2010-12-31 21:56:46 +00:00
  • 2bb7f6e821 If we're going to support the idea that things can log in without a body, we can't have the peer manager NPE when that happens. Par Winzell 2010-12-28 14:58:07 +00:00
  • faa7413d3c Nix unused import. Dave Hoover 2010-12-22 19:47:36 +00:00