Fixed more bugs. Now it's possible to generate a world without crashing.
Unfortunately, it's clear that there is a packet-sending loop going on.
I'll have to add a check to prevent integrated servers from spamming
themselves.
Finished implementing all the packet handling code. It could be improved
in the future to compress the initial packet sent to clients. With this,
the code is complete enough to run! Commands have not been fixed yet but
that will come in the future.
Threw out the complicated architecture that I'd made for synchronizing
server and client data perfectly. Instead, we now send just enough data
to the client and the resulting code is simpler. Some of the client-side
code is also done so all packet handling should be finished soon.
Continued building a system for transferring the complete internal
states of our dimensions from the server to the client. However, Steven
suggested that clients only need minimal data to operate properly, as
opposed to the server. My motivation for this more complicated system
was the concern that minimal information wouldn't be enough. I'm going
to commit my progress, then tear it down and write a much simpler
version.
Moved the DimLink code out of NewDimData in order to reduce clutter
inside that class and made it a separate class, except for functions
that should only be available for NewDimData. Deleted IDimLink and
changed all references to it to use DimLink instead. DimLink is now an
abstract class, which achieves the same encapsulation and protection we
had before by having DimLink implement IDimLink from within NewDimData.
NewDimData has a new class inside, InnerDimLink, which provides it
access to special functions that would be dangerous to expose. This is
the same mechanism used to protect NewDimData's dangerous functions.
These changes are in preparation for adding more code for packet
handling.
Not checking if a chunk existed before we placed a rift block could
cause the MC server to generate chunks in unloaded terrain. Bad things
commence. Possible world leak canidate.
Started rewriting our packet handling code. Deleted PacketHandler in
favor of using sided (Server-, Client-) packet handlers to make it
easier to follow what's going on in our code. Added some event-based
handling of updates which greatly simplified signaling that data needs
to be sent, but it's not completely done yet.
Modified how links are created so that the caller must specify a link
type when the link is created, rather than setting it later. This was
done to avoid having to send two link data packets following the way
links would be handled logically. Turns out this was good idea overall
for ensuring link integrity, because there was one case where I forgot
to set the link type after creating the link.
Fixed the code in DDTeleporter and made minor changes to other classes
that depended on those fixes. Ensured that PocketManager's load, save,
and unload methods are called appropriately and rewrote some of their
code. Made various changes in other classes (e.g. EventHookContainer,
PlayerRespawnTracker) to pass them references to DDProperties through
their constructors instead of having them rely on
DDProperties.instance() - this is a better programming practice in the
long run.
Renamed initialization methods in mod_pocketDim to make it clear that
they're called on events. Commented out command registration in
mod_pocketDim so that we can test DD as soon as PacketHandler is fixed,
without worrying about fixing the command classes.
Fixed the code in ItemLinkSignature and ItemStabilizedRiftSignature.
Removed obsolete code from PocketManager - we don't need to worry about
storing link keys anymore. All the data is stored in NBT, which means no
more generating unique keys or worrying about saving and loading them.
Moved tile entity classes to a separate package. Renamed some block
classes to match their in-game names (e.g. ChaosDoor -> UnstableDoor).
Moved TransientDoor to the blocks package. Cleaned up a little bit of
the code and automatically updated references to the classes that were
modified.
More cleaning up errors and code. There were a lot of things that needed
simplifying. Rather than fix the 3 or 5 copies of the same function
throughout the code, I made an effort to use the same function and
delete its copies.
Created the BaseItemDoor class to hold all the basic door item methods
that don't vary between types. That helped cut down on fixing things.
Also renamed the door item classes to match their in-game names.
There is still a ton of duplicate code out there.
Continued fixing things across various classes to make them work with
our new core classes. I've also cleaned up indentation and random code
snippets along the way.
1. Changed DungeonSchematic to use an external instance of Random rather
than initializing its own.
2. Created the Pair class, a strongly-typed tuple that was needed in
PocketBuilder.
3. Added a missing calculation in NewDimData for setting the packDepth
field during dungeon initialization.
4. Finished code missing in PocketBuilder. Changes listed above were
needed for this.
Replaced several core classes from DD with new classes to enforce
integrity checks. Rewriting everything that depended on those classes is
a massive undertaking but it should simplify our code and prevent the
many bugs we've seen lately. The rewrite isn't done yet, just committing
my progress so far.
Fixed the two blocks sticking out in the multilevel maze that made a
staircase unusable. Put a floor in simpleDropHall so it's actually
usable rather than killing the player.
DDLoot: Implemented a custom version of MC's generateChestContents() for
our own chests. It avoids two notable bugs that affect MC's version.
FillContainersOperation: Changed code to use
DDLoot.generateChestContents()
SchematicLoader: Fixed a bug in the way we calculated a seed for
selecting our dungeons that would cause certain seeds to dominate all
the others. Under certain circumstances, the function would only return
-1. That would make our dungeon selection severely biased. That was
resolved and the code was specifically tuned for seeding Java's Random
even for doors with nearly identical positions. The result was an
apparent major improvement in the randomness of dungeons.
ruins\rules.txt: Changed the dungeon generation rules to precisely match
the complicated scheme we had before. We're still using simple rules to
choose dungeons - I used a program to derive the effective distribution
of dungeon types that the old code would produce and converted it into
the current rule system.
Created NewLinkData and replaced references to the original LinkData.
Moved it to the mod_pocketDim.core package. Added Point4D, an immutable
point type for 3D integer coordinates with an added dimension ID.
Added code so that the mod loads dungeon packs stored in the custom
dungeons folder. Changed the code for loading bundled dungeons so that
they're loaded as dungeon packs. This means dungeon packs are now fully
integrated into the mod. The changes were tested and seem to be working
perfectly.
1. Integrated support for dungeon pack config options into the code
(i.e. we actually DO what the settings specify)
2. Added random transitions from one dungeon type to another. Dungeons
might also begin with a non-default pack.
3. Fixed a config reading bug that caused settings to be ignored and
some invalid settings wouldn't trigger exceptions. Also fixed other
dungeon pack bugs.
