- Separate FileMount into separate FileMount and WritableFileMount
classes. This separates the (relatively simple) read-only code from
the (soon to be even more complex) read/write code.
It also allows you to create read-only mounts which don't bother with
filesystem accounting, which is nice.
- Make openForWrite/openForAppend always return a SeekableFileHandle.
Appendable files still cannot be seeked within, but that check is now
done on the FS side.
- Refactor the various mount tests to live in test contract interfaces,
allowing us to reuse them between mounts.
- Clean up our error handling a little better. (Most) file-specific code
has been moved to FileMount, and ArchiveMount-derived classes now
throw correct path-localised exceptions.
- Add a new ClientJavaExec Gradle task, which is used for client-side
tests. This:
- Copies the exec spec from another JavaExec task.
- Sets some additional system properties to configure on gametest framework.
- Runs Java inside an X framebuffer (when available), meaning we
don't need to spin up a new window.
We also configure this task so that only one instance can run at
once, meaning we don't spawn multiple MC windows at once!
- Port our 1.16 client test framework to 1.19. This is mostly the same
as before, but screenshots no longer do a golden test: they /just/
write to a folder. Screenshots are compared manually afterwards.
This is still pretty brittle, and there's a lot of sleeps scattered
around in the code. It's not clear how well this will play on CI.
- Roll our own game test loader, rather than relying on the mod loader
to do it for us. This ensures that loading is consistent between
platforms (we already had to do some hacks for Forge) and makes it
easier to provide custom logic for loading client-only tests.
- Run several client tests (namely those involving monitor rendering)
against Sodium and Iris too. There's some nastiness here to set up
new Loom run configurations and automatically configure Iris to use
Complementary Shaders, but it's not too bad. These tests /don't/ run
on CI, so it doesn't need to be as reliable.
- Add a basic problem matcher for illuaminate errors.
- Add a script (tools/parse-reports.py) which parses the XML reports
generated by checkstyle and junit, extracts source locations, and
emits them in a manner which can be consumed by another set of
matchers.
This should make it a little easier to see problems for folks who just
rely on CI to test things (though also, please don't do this if you can
help it).
