Doesn't completely work because it checks the area again
and generates the chunks that were skipped over.
Signed-off-by: trivernis <trivernis@protonmail.com>
* Change generation to be asynchronous
At this stage it will most likely crash the server at a certain point.
Pausing and resuming isn't stable. Saving the progress isn't stable as well.
Chunks are being unloaded in the main thread by an unloader class.
* Switch to native threads
- Use thread instead of async tasks
- Store pending paper chunks in the database
- Interrupt the thread when it should be stopped
* Fix insertion of pending chunks
Fix an error that is thrown when the sql for inserting pending chunks doesn't have any chunks to insert.
* Add task states
Add states to differentiate between generating and validating tasks
as well as a field in the database to store this information.
A task will first generate a world until the required radius or the
worldborder is reached. Then it validates that each chunk has been generated.
* Add object representation of world_properties table
* Add DAO for pending_chunks table
* Add DAO for generation_tasks table
* Add state updating to periodic save
* Fix loading of world properties
* Add states to tasks and fix completion handling
* Fix progress report and spiral shape
* Modify the paper generation task so it works with spigot
This change is being made because normal chunk generation doesn't allow
chunks to be requested from a different thread. With PaperLib this issue
can be solved.
* Add workarounds for spigot problems
* Fix some blocking issues and update README
* Add locking to ChunkUnloader class
- Change the group to "Chunkmaster" to be able to disable all of those markers.
- Add a marker that displays the last chunk that has been generated and refreshes
every 30 seconds.
- Add getCenter command to get the center of the world used for generation
- Add setCenter to set the world center for generation tasks. If there is already a task running for the world the center for that task will not be affected.
- Fix SQL problem with multiple queries at the same time creating new connections
- Add translations for new commands
- Add Examples to Readme
- Add new table to safe world centers
- Add query to Generation manager to get the center of the world for new generation tasks
Fix error thrown by unique constrain for the world name in the sql table. The error occurs because the data from the database is loaded lazily after several seconds. If a task was started before that it wouldn't be checked against the existing tasks since they haven't been loaded yet. Fixed by loading the tasks if none exists when the generate command is invoked.
Add a delay to the starts of the tasks so that they don't all start on the same tick. To completely avoid two tasks being executed on the same tick, the "period" config option should be set to the expected number of tasks or higher.
- changed SqliteUpdateManager to SqliteManager
- added methods to execute sql with values and callback to the SqliteManager
- added auto close of statements
- switched to establishing a connection only when sql should be executed
- improved performance on paper servers by using chunk coordinates instead of chunk instances to check if the chunk was generated
- moved generation stuff to lib/generation
trivernis
trivernis
Trivernis