Replaces the jankson implementation with night-config. Night config is
what the forge api uses so it's easy to produce .toml files in the same
structure as CC:Tweaked.
All config options and comments from CC:Tweaked are implemented. Some
of these options are ignored in various bits of implementation but that
is another problem :)
Also splits the config into a client and server file. The server file is
saved per-world. Names and locations are consistent with CC:T.
Adds a sidebar to the computer and turtle GUI. This currently provides
- A power indicator, which turns on/shuts down a computer.
- Button to queue a "terminate" event
I've written three or four integrations with bundled cables now over the
various versions, and I'm still not convinced this one is right. It
appears to work though, so we'll see.
We need depend on the main jar here (rather than the API jar), as
otherwise our javadoc tools fails - the API references some internal
classes which aren't on our classpath.
It might be nice to write tests for this (and CT) in the future -
goodness knows these features are fragile - but that'll require some
more gradle work which I don't want to do right now.
Closes#772
- Move registry code into the various *Registry classes.
I'm not sure this is any more sensible, but things being registered
in different places kinda irked me.
- Everything else (i.e. event listeners) goes in a {Client,Common}Hooks
class right now. It's not ideal, but I don't think we can split it up
much.
I refactored the mixin package to be within a fabric package so other
fabric specific code could be kept nearby. But a mixin from a recent PR
got left behind in the merge. Github's empty merge conflict might have
been about this.
Made a top-level fabric package so mixins and their ancillary classes
can be grouped together. Also removed a stub duck interface that got
left behind somewhere along the line (MixedFirstPersonRenderer).
Forge's ITileEntity interface adds a onChunkUnloaded handler method into
block entities. The fabric port doesn't have a re-implementation of this
feature, which meant TileMonitors weren't releasing their buffers when
unloaded. This commit adds that handler method back into TileGeneric,
which all CC block entities inherit from. Handler logic for the four
block entities that use this feature were copied over from the forge
repo.
These buffers were being released on the server thread, causing the
logical server in a single player session to crash just before shutting
down. It seems the crash happened late enough to not cause world
corruption issues, and so went unnoticed. It was however causing a
memory leak when quitting/rejoining worlds, and it left the save file
lock in limbo.
- Fix double updateOutput() call in TileComputerBase - I guess a
merge/rebase gone wrong in the past.
- Don't call updateBlock() when creating a server computer. This used
to be needed when we sent the computer to the client, but this is no
longer the case.
- Don't call updateBlock() on TileMonitors when updating from the
client. We don't need to do a redraw here, as this is all stored in
the block state now.
- Don't update the block when reading turtle upgrades. See #643 for
some background here.
See #658
- Return a more sensible string for empty treasure disks (i.e. those
given by /give). This should help identify packs which are giving
items in non-supported ways.
- Fix NPE when the treasure mount doesn't exist.
Fixes#801
Implementation is a little awkward, as we can't send OPEN_FILE links
from the server, so we ensure the client runs a
/computercraft open-computer ID command instead. We then intercept this
on the client side and use that to open the folder.
