Repository navigation
Conversation
Defer frame buffers and metrics until after the BWAPI handshake, and use a focused Windows shared-memory mapping interface. Automatically load an architecture-named JNA DLL beside the bot JAR after verifying it against the embedded binary. Preserve explicit JNA settings and ordinary loading when no matching bundle is available, without a custom native cache. Package portable Windows DLLs beside the JAR and cover native loading, startup ordering, mapping behavior, and Windows Java 8 x86 in tests and CI.
| client.reconnect(); | ||
|
|
||
| // Allocate game bookkeeping and async frame buffers only after the BWAPI handshake. | ||
| this.performanceMetrics = new PerformanceMetrics(configuration); |
Collaborator
There was a problem hiding this comment.
My naive assumption is that change is actually the most important for the startup boost (e.g. the Client starting to listen ASAP). Do you have individual timing on how much each of the 3 changes actually brings individually?
|
|
||
| Client(BWClient bwClient) { | ||
| this.bwClient = bwClient; | ||
| JnaNativeLoader.initialize(); |
Collaborator
There was a problem hiding this comment.
nit: as we are checking the windowsOs one line below and this optimization is for windows anyways, perhaps we can transform this in a simple if else instead?
if (System.getProperty("os.name").toLowerCase().contains("win")) {
JnaNativeLoader.initialize();
clientConnector = new ClientConnectionW32();
} else {
clientConnector = new ClientConnectionPosix();
}| /** | ||
| * Opens the BWAPI server's game table and game-state shared-memory mappings. | ||
| * Binding only these two kernel32 functions avoids initializing JNA's full | ||
| * Kernel32 proxy before the client can discover a server. |
Collaborator
There was a problem hiding this comment.
Please add a small comment on how much time this is expected to save
| SynchronizationEnvironment environment = new SynchronizationEnvironment(config); | ||
| environment.runGame(20); | ||
| final double minObserved = 0.25; | ||
| assertTrue("Copy to buffer: minimum must be positive", |
Collaborator
There was a problem hiding this comment.
I copied your fix for broken head test in #101, thanks! Please rebase on top of develop head
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Most software stacks that run client bots (SC-Docker, Tournament Manager, SCHNAIL, for example) don't bother checking whether the client is connected before starting a game. There's not even an obviously good way to do this.
Instead they just launch the client bot, wait a few seconds, then launch StarCraft and pray the client is ready. This usually works, but when it doesn't, the bot fails to connect. The game is ruined.
This pull request aims to prevent failed connections by reducing the time required for client bots to be ready for connection. The measures include:
These changes were vibecoded but have undergone several rounds of iteration and testing. They reduce time-to-connection-ready by about 73%.