this was making the location table point to an offset that did not yet exist, which caused the region header consistency check to discard the region as corrupted the next time it was loaded.
this works around a bug where corrupted text on preexisting signs can mess up the client. This also prevents corrupted text getting onto signs in the future by having them scrubbed and validated before applying them.
this fixes effect durations being off (mostly), closes#2650
there are still some minor differences, but this is closer matching than the previous version.
this is a partial fix for #2717, but still not ideal because it'll spam whenever a chunk is attempted to be generated. However, fixing this properly requires potentially breaking API changes.
this now has a mouthful of a name. I'd like to invert it, but I can't do that without silently breaking backwards compatibility, which is unacceptable.
not ready to call this "fixed" yet because any chests that were already affected by the bug will still be affected. This change will prevent the creation of more broken chests like this.
Plugins would indirectly trigger permissible recalculation too early in the login sequence, which then caused their permissions to be recalculated and subscribing them to the broadcast permission far too early.
generating 1 large bounded random costs the same as generating 4 small ones, so it makes more sense to do it like this instead.
Note that prior to 7.1 this code would not work due to it not handling 64-bit appropriately.
In vanilla it doesn't drop the exact number of points you collected. Rather, you lose a little for every level above 1 you had (1 level requires 7 points, later levels require +2 per level), and can recover at most 100 points. Hence, if you had 10 levels, you get back enough points to fill 5 levels and most of a 6th. 14-15 levels gets you the upper bound of about 7.5 levels.
This is not identical to vanilla, but I don't care because it gets rid of edge cases and also makes it easier to integrate with EntityDeathEvent in the future.
I believe this is caused by a bug in the linux kernel, since it only impacts certain machines I tested (one, to be specific). Whatever the case, setting a max backlog size is prudent anyway, and fixes the problem.
Packets with a too-short payload would either cause the RCON thread to hang until the client disconnected, or crash the RCON thread entirely.
commit 90bb1894d7f87645b806f5fc67d1b877bb963180
Author: Dylan K. Taylor <odigiman@gmail.com>
Date: Tue Jan 22 18:15:46 2019 +0000
fix some bugs in RCON
This will now throw an exception at the source instead of crashing when the entity is saved, which should put the blame on the correct plugin responsible for this.
This also includes magic method hacks to preserve backwards compatibility, since the fireTicks field is now protected.
Encoded tags larger than 32KB overflow the length field, so we can't send these over network. However, it's unreasonable to randomly throw this burden off onto users by crashing their servers, so the next best solution is to just not send the NBT. This is also not an ideal solution (books and the like with too-large tags won't work on the client side) but it's better than crashing the server or client due to a protocol bug. Mojang have confirmed this will be resolved by a future MCPE release, so we'll just work around this problem until then.
If a plugin was involved in a crash, we can't safely blame it for the crash, since it might have innocently triggered a core bug. Furthermore, it's difficult to accurately detect plugin causing things like invalid argument crashes.
it's much less expensive to just calculate the modulo of the current tick and 20, and overwrite past entries. The effect is the same. The only difference is that the arrays won't be ordered by time, but that doesn't matter anyway.
This is better for performance because these then don't need to be reevaluated every time they are called.
When encountering an unqualified function or constant reference, PHP will first try to locate a symbol in the current namespace by that name, and then fall back to the global namespace.
This short-circuits the check, which has substantial performance effects in some cases - in particular, ord(), chr() and strlen() show ~1500x faster calls when they are fully qualified.
However, this doesn't mean that PM is getting a massive amount faster. In real world terms, this translates to about 10-15% performance improvement.
But before anyone gets excited, you should know that the CodeOptimizer in the PreProcessor repo has been applying fully-qualified symbol optimizations to Jenkins builds for years, which is one of the reasons why Jenkins builds have better performance than home-built or source installations.
We're choosing to do this for the sake of future SafePHP integration and also to be able to get rid of the buggy CodeOptimizer, so that phar and source are more consistent.
Since 3.2 there have been some runtime performance issues related to garbage collection and dynamic AsyncWorker booting. This is partly because of GC being dumb about shutting down what it thinks are "unused" workers. A worker which has been idle for a single tick is considered the same as a worker which has been idle for hours. The result of this is that on active servers, workers would get shut down and then immediately restarted because of something like chunk sending. Since booting an async worker is frightfully expensive, this causes lag spikes, which is obviously bad.
This commit changes the GC mechanism to only shutdown workers which have not been used for the last 5 minutes.
This was removed way back in 2016 because of an unidentified bug which caused permissible commands not to show up on the client. Back then, command parsing and validity checks were client-sided, and the client would simply not send the command at all if it didn't recognize it. Now, that problem is gone, so it doesn't matter as much if there are permission bugs which cause commands to be erroneously missing.
closes#2625
this bug has existed for so long I forgot it was still here. People stopped pestering me to do something about it, and as a result I forgot to do anything about it.
This hack isn't perfect, but it filters out the worst of the noise. It has side effects for legitimate fast double-clicks, but I don't think anyone will be too bothered - just click more slowly.
This hack may also have negative side effects on poor connections where latency spikes are a problem, but there isn't really much that can be done about that.
Always order chunks ASAP on chunk change, not just during the spawn sequence. This fixes the sluggishness observed in BlockSniper when doing async chunk modifications.
I don't know why this would be missing, but in some cases it is, as seen in the crash archive. Whatever the case, we shouldn't be shitting the bed because of this.
this was observable by planting a sapling underneath an existing tree and punching it with bone meal.
This change will also prevent trees generating too close together.
this should have been done a long time ago, but we didn't want to cause compatibility problems with CA. Now it enforces version checks, this isn't a problem anymore.
this doesn't fix shit but it at least doesn't crash. Fixing this properly can't be effectively done any other way without backwards compatibility breaks. Fortunately it's not common practice to grow trees at the top of the world.
this attribute is not visible on the client and is only used for controlling saturation depletion. It's extremely spammy and as such really shouldn't be sent over network. This has also been causing some minor client-side performance issues in survival.
this must have been missing for how many years now? thanks @shoghicp
this is why we don't do releases on friday night... in my defence my device had the beta installed...
This allows to save the language without rewriting pocketmine.yml. Since this is a "standard" config option (something that the user might want to directly modify) it's reasonable to put it in server.properties. pocketmine.yml is generally reserved for more advanced configuration options.
this technically involves non-breaking API changes which should happen on a patch release, but I can't be bothered with the dust cleanup, so we'll just blow it away now. It doesn't hurt anyone anyway.
This was the cause of a bug with regeneration which caused players taking fatal damage under regeneration not to die correctly. On the server side they would die and immediately regenerate some health, which would cause the next attribute sync to not report the health drop to zero, which made the client unaware that it was dead.
Perhaps attributes should be forcibly synced in some circumstances, but nonetheless regeneration shouldn't apply post-death.
This is an incremental improvement over 4a6841a5a4. This change works better because it also reduces disk spam of crashdumps.
This will now sleep if the server uptime was less than 120 seconds before crashing. If unattended, this will clamp down on automated crashdump spam. If attended, the user can simply press CTRL+C to abort the process and skip the delay.
This takes advantage of two key behaviours of PHP:
1. Assigning a string does not copy the string
2. Changing an offset in a string causes the string to be copied.
These two factors combined, along with the fact that blocklight and skylight arrays are usually all-zeros, allow us to produce a significant memory usage reduction of loaded chunks.
A freshly generated PM world with 3,332 chunks loaded drops from 310MB to 200MB memory usage with these changes applied.
This was suggested recently by @TheDeibo. We don't want users running source-code installations unless they are developers, and developers should know how to boot a source-code installation anyway.
this process of fast-serialization, fast-deserialize, network-serialize is an order of magnitude slower than just doing the network encode directly on the main thread, and also copies more useless data.
For the main thread, the figures were something like 3x more expensive, and then an extra 7x for deserialization on the worker thread. This is a ridiculously large overhead.
this is perfectly fine to use, and preferable to getting a cyclic ref to the scheduler. TaskScheduler->cancelTask() does pretty much the exact same thing, and the scheduler internals are designed to deal with this anyway.
this should produce some reduction in spam at the source.
This could also be used to control the rate at which constantly-crashing servers restart to stop them spamming the disk as well, but the main concern here is eliminating crash archive involuntary DDoS by crashy servers.
This supersedes addChunkPacket() in most cases, and has a more clear name. It broadcasts the given packet to every player who has the target position within their chunk load radius.
This can happen when a light source is removed and later encountering another light source to fill the gap. A higher light level may get set and then not propagated. This bug is difficult to explain, but fairly easy to reproduce.
Grass can cause issues here by requesting blocks randomly offset away from itself, which can cause silent chunk loading on chunk ticking. It also causes crashes if chunk autoloading is taken away, which is obviously undesired.
It was also noticed that player chunkloaders cause chunks to start getting ticked as soon as they load their first chunk, which is before the entity is visible to everyone else on the server. This is probably undesired behaviour.
this can happen when eval() is used, and then we get a big blank mess with nothing on it. eval() is a special case that should be handled separately, but for now this is just fixing a bug.
This was observed in a recent crashdump where a plugin triggered a recursion error, but the stack trace did not contain any sign of a recursive event call. I conclude that this must have been caused by previous event handlers triggering errors 50 times in order to make the recursion detection break, because the recursion detection did not decrement the counter in cases where an exception was thrown.
while we can't deal with this information, it's needed for the sake of unit testing so we don't shit on every bit of incoming data of these packet types.
Use of this by plugins will produce a lot of undefined behaviour, such as event handlers not being unregistered, scheduled tasks not being removed, and registered permissions causing memory leaks.
This check is completely unnecessary since handlers get unregistered when a plugin is disabled. Additionally, this is an extremely hot path and this change produces a modest 5% performance improvement to event calls.
this is cleaner than leaving the player hanging for 5 seconds (which they'll often timeout from anyway). Banning the IP without kicking the player can often look like "lag" and end up getting brushed off as a performance issue.
this dates back to the days where PM used to kill threads to stop them. Today we're more civilized and ask it to stop nicely, so this isn't necessary anymore.
This has the triple bonus effect of a) making a lot of code easier to read, b) reducing Server::getInstance() usages, and c) removing a whole bunch of Server dependencies.
The network and block namespaces are untouched by this commit due to potential for merge conflicts. These should be dealt with separately on master.
This is dependent on the changes made in b1e0f82cbf. This now makes it possible to call events without fetching a Server reference, allowing to eliminate a vast array of Server dependencies.
The basic principle here is "if you're not expecting it, don't catch it".
Event handlers are **never** supposed to throw exceptions. If they do throw exceptions, it's usually going to one of two things;
1. Broken code producing an error
2. Code triggering (and not catching) a runtime error
Both 1) and 2) boil down to defective code on the part of the event handler, and thus should not be caught by the caller, but instead allowed to crash the server and produce a crashdump.
It's also undesirable to catch unexpected errors here for a few other reasons
- It leaves the owner of the event handler in an unknown, potentially unstable state
- It allows broken code to cause event handlers to spam the logger in events that happen frequently (for example movement handlers)
- It allows the process to continue down a train of further undefined behaviour, which may lead to more errors or ultimately a crash, so it makes no sense to hold off the inevitable.
This has a few advantages that are not merely inverted disadvantages:
- Crash dumps will now be created and automatically submitted for defective event handlers, allowing quicker issue location, debugging and fixing in plugins without manual user interaction
- Event calling now isn't dependent on Server to work.
This assumes that the region is properly garbage-collected and packed, but if the file contains uncollected garbage this may not be the case, resulting in a region larger than a gigabyte.
readline on Windows causes issues with console output corruption. Additionally, PM readline impl is extremely buggy and probably ought to be removed. However, have a hotfix for now.
This came to light after observing cfb6856634 in a fresh light. I noticed that this fix should not have been necessary because clearPermissions() should have dealt with it. Unfortunately, permissions can be set without being set in PermissibleBase->permissions, so this misses things.
This is needed for batched lighting updates to work. It also reduces the overhead involved with simply preparing a lighting update and moves the pain to the execute() instead.
this usually causes the console to get spammed with errors. Additionally, in the case where doTick() throws any exception, it's usually because we're in a state we didn't want to be in, so we really should not carry on trying to keep ticking when something breaks here. Instead, this should generate a crashdump.
Previously every thread using the logger had to inherit runtime-defined INI entries in order for the timezone to be set correctly. This removes that requirement.
- Don't allow the same window ID to be used when another window is already using it
- Detect window ID collisions when selecting IDs for regular containers (should never happen, but anything is possible)
this is an absurd bug that nobody would ever otherwise notice, but the problem is that the doOnFireTick() call isn't evaluated if hasUpdate is already true.
leaving out turtle helmet for now because of complications relating to the effect application - I REALLY don't want to tick armour if I can avoid it, due to the performance concerns.
A common pitfall developers fall into with this function is that it has to be called from the scope of the tile class you're creating NBT for, but people commonly do Tile::createNBT() directly, which then results in cryptic "Tile is not registered" errors. This now throws a BadMethodCallException instead to be fully clear about this.
In the future this will be removed completely once NBT is no longer required to create a tile, but for now this is a confusing issue that should be dealt with.
PhpStorm can't see constructor usages when the class name is dynamic. This causes maintenance problems because cross-referencing constructors called like this doesn't show up dynamic calls.
I am eating my own words this once, because having the tester plugin as a separate repository makes no sense - it is just added barriers to writing proper tests with no actual benefit. Since the tester plugin is specifically intended for CI, it doesn't make sense for it to be in its own module.
The original regex almost completely failed at its objective, because it a) only worked if there was no value for the key, and b) did not prevent all such occurrences getting transformed, while quoting patterns that would not get transformed anyway.
There's no implementation here yet, but that can come later. This lays the ground for allowing plugins to have an integrated method to send forms, as well as a solution to the ID conflict problem.
A built in implementation should not be a concretion and it should be able to be swapped for third party implementations. This enables the possiblity to do so.
Previously any random error could occur during an AsyncTask preparing a chunk, and the Level would never know about it and thus never send the chunk.
I don't know how many invisible-chunk bug cases this fixes, but I expect it's quite a lot.
RuntimeException is very generic and might be thrown for other reasons apart from web request failures.
This is backwards compatible because InternetException is a descendent of RuntimeException. Additionally, getURL() and postURL() have intentionally been left untouched for backwards compatibility's sake.
* Added a new PermissionManager, remove ridiculous cyclic dependencies of Permissions on Server
Aside from all the other ridiculous design problems with the permission system, the biggest problems are its API. This is, once again, a result of poor API design copied from Bukkit.
This pull request removes all permission-related functionality from `PluginManager` and moves it to the `pocketmine\permission\PermissionManager` class.
As can be observed from the removed code in the diff, the permissions system was previously entirely dependent on the Server, because it needed to get the PluginManager for registering permissions. This is utterly ridiculous. This refactor isolates _most_ permission-related functionality within the `permission` namespace.
As mentioned above, this stupid API is a direct result of copying from Bukkit. If you look at the API documentation for Bukkit for `PluginManager` you will see that the methods I'm deprecating here are also in there.
## Changes
- Added a new `PermissionManager` class. This can be accessed via its singleton `getInstance()` static method.
- Deprecated the following `PluginManager` methods - these will be removed no later than 4.0.0:
- `getPermission()`
- `addPermission()`
- `removePermission()`
- `getDefaultPermissions()`
- `recalculatePermissionDefaults()`
- `subscribeToPermission()`
- `unsubscribeFromPermission()`
- `getPermissionSubscriptions()`
- `subscribeToDefaultPerms()`
- `unsubscribeFromDefaultPerms()`
- `getDefaultPermSubscriptions()`
- `getPermissions()`
- Added `Internet::getIP()`, `Internet::getURL()`, `Internet::postURL()`, and `Internet::simpleCurl()`.
- Deprecated the corresponding functions in `Utils`. Updating to the new functions is as simple as replacing `Utils` with `Internet`, since this doesn't break backwards compatibility.
The deprecations should be catered for by plugin developers. These deprecated redirects will be removed no later than 4.0.0.
Under normal circumstances, none of the boxed code will throw exceptions. Under exceptional circumstances, the caller should know about it. Usually the caller is the server. We don't want to catch unexpected exceptions because those should crash the server and generate a crashdump.
As discussed in #2297:
Honestly I don't see a fit purpose for async saving at all. It should either always be synchronous or always asynchronous, and at the user's own option. However, this isn't currently possible because Config doesn't enable you to get the serialized content without writing it to disk.
Consider the following code:
```php
for($i = 0, $size = $this->getServer()->getAsyncPool()->getSize(); $i < $size; ++$i){
$this->getServer()->getAsyncPool()->submitTask(new class extends AsyncTask{
public function onRun(){
sleep(5);
}
});
}
$config = $this->getConfig();
$config->set("steve", "hi");
$config->save(true);
$config->set("steve", "bye");
$config->save(false);
```
Output:
```yml
---
steve: hi
...
```
Expected output:
```yml
---
steve: bye
...
```
Additionally, if your configs are causing you performance issues when you're saving, it's a clear sign that
a) you're saving too much
b) you're abusing configs and should consider using a database.
Configs should be used for _simple_ data which does not change much. Configuration is such that the _user_ is expected to be able to modify it. As such, it should never be an issue to save synchronously.
In the future, something like ReactPHP may be introduced to allow proper async saving. When this happens, async saving would always be sequential but non blocking. Using threads for this makes no sense.
I wasn't sure whether this would be considered a bug fix or a feature. Nonetheless, it's a behavioural change, so it belongs in 3.1 if anywhere.
Prior to this, plugins would be required to call saveDefaultConfig() before calling getConfig() or anything else. Calling getConfig() without saveDefaultConfig() first would generate an empty configuration file. Instead, it now saves the default config before loading it.
this goes on 3.1 because it changes the behaviour of chunk cloning, which might possibly break some plugins, and this isn't a bug fix.
This should see no change in behaviour other than a minor performance improvement and slight reduction in memory usage.
This allows controlling how arrows are picked up:
- by anything
- by only creative players
- by nothing
This adds new API methods to Arrow:
- getPickupMode()
- setPickupMode()
This adds new public constants to Arrow:
- PICKUP_NONE
- PICKUP_ANY
- PICKUP_CREATIVE
//If PHP version isn't high enough, anything below might break, so don't bother checking it.
return[
\pocketmine\NAME." requires PHP >= ".MIN_PHP_VERSION.", but you have PHP ".PHP_VERSION ."."
];
}
}
if(extension_loaded("pthreads")){
$pthreads_version=phpversion("pthreads");
if(substr_count($pthreads_version,".")<2){
$pthreads_version="0.$pthreads_version";
$messages=[];
if(PHP_INT_SIZE<8){
$messages[]="Running ".\pocketmine\NAME." with 32-bit systems/PHP is no longer supported. Please upgrade to a 64-bit system, or use a 64-bit PHP binary if this is a 64-bit system.";
critical_error("Selected PHP binary ($binary) does not satisfy some requirements.");
foreach($messagesas$m){
echo" - $m".PHP_EOL;
}
critical_error("Please recompile PHP with the needed configuration, or refer to the installation instructions at http://pmmp.rtfd.io/en/rtfd/installation.html.");
$this->getLogger()->debug(sprintf("Level \"%s\" took %gms, setting tick rate to %d ticks",$level->getName(),(int)round($tickMs,2),$level->getTickRate()));
$this->getLogger()->debug(sprintf("Level \"%s\" took %gms, setting tick rate to %d ticks",$level->getName(),(int)round($tickMs,2),$level->getTickRate()));
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.