This change subtly broke the poll() loop when a connection was removed,
because erasing an entry from fds would invalidate the iterator that
was still being used.
This reverts commit ec67cc6527.
CNProtocol, CNShared, CNStructs and Defines are now in core/.
CNLoginServer, CNShardServer and Monitor are now in servers/.
core/Core.hpp wraps all the core headers except for CNShared.hpp.
Defines.cpp has been renamed to Packets.cpp, and so has its
corresponding namespace, but not the header file. This is in preparation
for upcoming changes.
ChatManager -> Chat
MissionManager -> Missions
NanoManager -> Nanos
TransportManager -> Transport
ChunkManager -> Chunking
BuddyManager -> Buddies
GroupManager -> Groups
RacingManager -> Racing
ItemManager -> Items
NPCManager and PlayerManager remain.
Note: You can use git log --follow src/file.cpp to trace the history of
a file from before it was renamed.
All packet handlers and helper functions that are only used in the
source file they're declared in have been taken out of the namespaces in
the corresponding header files, have been marked static, and have been
reordered to avoid the need for declarations at the top of each source
file.
Each source file now contains a "using namespace" directive so that the
static functions don't need to prefix the source file's symbols with
their namespace. All redundant namespace prefixes found have been
removed.
An unused nano power resetting function in NanoManager has been removed.
* Database.hpp is still the only external include file (moved to db/)
* The header is still uppercase to match its namespace
* db/internal.hpp is the shared header for the DB source files
* Added -Isrc/ compile flag for src-relative include paths
* Hoisted CHDR above CSRC in Makefile (it was bothering me)
* make clean now removes all objects in the subdirectories as well
This is terrible. It was a mistake to do this before cleaning up the
actual code. It might be better not to use this commit and to do this
refactor in a different order or something.
I've kept all the functions in their original namespaces for now, since
putting them all into the same one will cause collissions, and this is
all getting rewritten soon anyway.
* Extracted all commands into CustomCommands.cpp
* Moved all chat-related packet handlers into ChatManger.cpp
* Cleaned up redundant includes
* Unified handler naming scheme
* Made all command handlers in CustomCommands.cpp static
Also reject players requesting more than 6 missions.
This is just a minimal measure to prevent replaying mission completion
packets. This part of the codebase will be largely refactored soon, so
more through changes can wait.
In certain circumstances, like when running a private server through
Hamachi, the shard IP will be set to an address the local machine can't
reach itself from, preventing only the local player from getting past
character selection. This workaround detects local connections and
sends a loopback address for the shard instead of the configured one.
This makes those use cases feasible.
Also reorder the rapid fire check in MobManager::pcAttackNpcs(), so the
output packet validation happens immediately before the buffer is
initialized, for clarity.
* Create new table to store redeemed codes
* Check if a player already used a code when using /redeem
* Change Coordinate columns to non-plural form
* Fixed EmailItems unique constraint not being specific enough
* Bumped DB version to 3
This is important because the client can genuinely send such an invalid
packet by mistake during normal gameplay.
If a sanity check fails, we don't need to send any sort of "move it but
keep it where it is" packet, since simply ignoring the invalid request
doesn't softlock the client.
Also improved validation of inventory slot indexes.
* Muting a player's freechat
* Kicking players
* Querying info about a player
* Teleporting yourself to a player
* Teleporting a player to yourself
* Teleporting a player to another player
* Teleporting a player to arbitrary coords
* Teleporting a player to arbitrary coords in an arbitrary mapnum
* /unstick
Also:
* Renamed misleading setSpecialPlayer() to setValuePlayer()
* Revamped monitor logic
* Added server-side checks to account level 50 commands
* Made sure even trade chat is run through sanitizeText()
* Moved setSpecialState() closer to its calling functions
* Interpret client commands even in Buddy and Group chat (but not in
Trade chat)
- Battery drain does not set your battery below 0.
- Debuffs cast by enemies take into account intensity now. (Snare will work)
- Used int_max to replace arbitrary 20000, added more comments.
- combatStep was altered, the mob can attack and give chase at the same time. Kiting melee mobs around is much harder.
- Mobs in general are more harder, closer to how it was in retro.
- Nanos styles are less detrimental to battle.
* Do not define an sqlite_stmt* named stmt in a loop such that it shadows
an outer variable of the same name.
* Do not re-prepare a statement that has been reset.
* Fixed not being able to modify the DB externally while the script is
running
* Made most DB-related errors print the appropriate error string
* Fixed the settings looking for dbsaveinterval in the shard category
instead of the login category
* Added -shr and -wal to .gitignore (even though we're not actually
using WAL mode)
Hopefully the DB code is now free of resource leaks and undefined
behaviour.
This fixes the crash with mobs with a very small m_iIdleRange and avoids
unnecessary looping.
Co-authored-by: JadeShrineMaiden <69916714+JadeShrineMaiden@users.noreply.github.com>
* Respawn the player in the air, not on the ground
* Do save the player's current instance, just to lessen the chance of
validation not catching stale recall coords.
* Single quotes for strings
* Semicolons at the end of all commands
* No double colons around identifiers
* Spaces for alignment
* Some parts indented for readability
Not everything is perfectly consistent, but it should all be pretty
readable now.
Non-SQL changes:
* Opening braces shouldn't be on their own line
* Removed trailing spaces
* exit(1) should be used instead of terminate(0) during server init
implemented getCharInfo
implemented delete character
implemented evaluateCustomName
implemented name change function
implemented getPlayer function
implemented functions for buddies
implemented emails
implemented addAccount
implemented updateSelected
implemented findAccount validateCharacter isNameFree and isSlotFree
implemented getTableSize function
implemented functions for name save, character creation and finishing tutorial
- Eruption is now blocked by stun and sleep.
- Corruption should block all nano abilities.
- Buffs time out for other players
- Timed mission bugfixes (AGAIN)
- Corruption and Eruptions fire quicker.
- Heal egg ids fix
- No power nanos no longer break the system.
- Mobs should no longer restun.
- Mob ability chance calculation adjustments.
- Duration of the power's debuff is sent as iDamage instead of 0, this removes the ugly "Block" that shows up on successful hits.
- Group mob respawning bugfixes
- a bit of a cleanup
Recheck the entry at the current index after we remove one. This isn't
strictly necessary, since the next pass will get it anyway.
Using a vector as opposed to our own realloc()'d array means indexing
beyond the size() is undefined behavior, so it's better to be safe.
This is how it should have been handled anyway. This fixes a (benign)
race condition when the server is killed, as well as gprof builds.
Also updated tdata.
This fixes a crash due to the invalidated iterator continuing to loop.
It should be safe to assume there's only ever one duplicate/stale player.
Also updated tdata.
* Fixed Nano stamina not being halved on respawn
* Reverted the default argument to terminate() change because MSVC is
undable to disambiguate the function pointer passed to sigaction()
* Fatal errors during init (like in TableData) can just call exit(1)
directly anyway (missing "OpenFusion: terminated." be damned)
* Switched to a slightly more portable syntax for getting the version
in the Makefile
* We shouldn't join the shard thread in the signal handler because the
thread the signal handler ends up running in is undefined behaviour and
we don't strictly need to join it anyway
Many of these issues were discovered on OpenBSD.
* Gave it a default argument, since we never actually care about it, but
it needs to have it to conform to the signal handler prototype
* Constricted the area locked by activeCrit to only the block that deals
with the connections vector, to lower the chance of a future badly
placed call to terminate() deadlocking the server instead
* Extracted PollFD manipulation and nonblocking socket configuration
into helper functions
* Replaced the connections list with an unordered_map
* Dynamically grow the number of PollFD structures with realloc()
With these changes done, the server's CPU usage is completely diminished
from its old average of ~47% to ~0.07%, with occasional spikes up to ~14%.
This change required sacrificing both code cleanliness and the specific
error reports from perror(). Those could have been kept with a portable
wrapper, but that's too much work. We'll do it if unforseen errors
arise.
