godruoyi / php-snowflake Goto Github PK

设置不同的开始时间，生成的ID位置可能是18位，可能是19位，而且看某些平台都是1开头的19位，这是在18位上加了一个前导1？

see https://phpstan.org/

reference

• https://github.com/laravel/prompts
• https://github.com/laravel/folio

用 ab -c 200 -n 10000 http://127.0.0.1:18306/test/id
至少有三分之一的重复ID生成.老兄这是怎么回事?

53bit is enough for my project
pls let me know, thanks!

时钟回拨除了抛出异常外，能有其他的解决办法吗？

datacenter AND workerid will be NULL，Will cause duplicate id to be generated in the for loop

DEFAULT HANDLER - Error(code:0): Class 'Godruoyi\Snowflake\swoole_lock' not found

According to the suggestion in #53, add perform concurrent testing for the following resolvers below.

RedisResolver/FileLockResolver/SwooleResolver

As you probably know, PHP no longer actively supports versions below 8.1 (reference). I wonder if you would be open to dropping support for everything lower than 8.1 and releasing a new major version with PHP built-in type-hints added to every method and interface.

As this is a major change it would require a new version release (3.x). With this change, we could bundle other breaking changes if you are looking into doing any of them.

Hello!

I'm trying to integrate this into my Laravel application and I had some questions surrounding setting the starting timestamp:

• First and foremost, why would the starting timestamp be set to anything other than the Unix epoch (1970-01-01 00:00:00 Z)? This would be the most intuitive date and is generally what a timestamp is assumed to be based on.
• Should I set my own starting timestamp or just leave the default? What are the implication of one over the other?
• It seems like the default timestamp use is 2019-08-08 08:08:08 which is seemingly random?

Thanks!

$snowflake = new \Godruoyi\Snowflake\Snowflake;

echo $snowflake->id() % 32; //0

这里打印的timestamp，每个几秒，都变成间隔了几个小时，看看是不是bug，还是我理解错误了，并不是时间？

    $temp = $snowflake->parseId($id,true);
    echo "time:",date("Y-m-d H:i:s",($temp["timestamp"])),"\n";

可以设置长度吗？生成的编号大于第三方接口最大长度，所以想减少2位

一个奇怪的问题会生成负数，2台电脑一台不会生成负数，一台很大几率会产生负数

Thanks for this package!

I'm just trying it out with Laravel and I get this when I use the attach() function:
SQLSTATE[HY000]: General error: 1364 Field 'id' doesn't have a default value

Any ideas? Thank you!

假设有users, orders两张数据表，通过snowflake来为他们生成唯一主键id.

如果需要使这两种id的sequence不同怎么处理比较好？

目前想法是实例化两个不同的单例，每个RedisSequenceResolver的前缀设置成不一样的。
不知道有没有更好的方法。

示例:

$this->app->singleton('userIdGenerator', function ($app) {
    return (new Snowflake(0, 0))
        ->setStartTimeStamp(strtotime('2022-07-20')*1000)
        ->setSequenceResolver(
            (new RedisSequenceResolver($app->get('redis')))->setCachePrefix('UserId')
        );
});

$this->app->singleton('orderIdGenerator', function ($app) {
    return (new Snowflake(0, 0))
        ->setStartTimeStamp(strtotime('2022-07-20')*1000)
        ->setSequenceResolver(
            (new RedisSequenceResolver($app->get('redis')))->setCachePrefix('OrderId')
        );
});

能否增加索尼雪花算法,分别是 1-39-8-16

寿命（174年）长于雪花（69年）
它可以在比Snowflake（2 ^ 10）更多的分布式计算机（2 ^ 16）中工作
它可以在单个机器/线程中每10毫秒最多生成2 ^ 8个ID（比Snowflake慢）

go语言的实现：https://github.com/sony/sonyflake

getCurrentMicrotime in 1 second may return samme value , id() functino while loop usleep(1) have meaning？

Hello everyone!

I'm currently testing Snowflake to implement it in an API I'm working on for generating unique IDs. I have the following script for running some tests and validating that the IDs are not duplicated.

$processes = 10; // Number of processes to create
$iterations = 1000; // Number of iterations per process

$ids = [];

// Function that generates IDs and checks for duplicates
$generateIds = function () use (&$ids, $iterations) {
    for ($i = 0; $i < $iterations; $i++) {
        $snowflake = App::make('snowflake');

        $id = $snowflake->id();

        if (in_array($id, $ids)) {
            echo "Duplicate ID found: $id\n";
        } else {
            $ids[] = $id;
        }
    }
};

// Create processes and execute the ID generation function in parallel
$childProcesses = [];
for ($i = 0; $i < $processes; $i++) {
    $pid = pcntl_fork();

    if ($pid == -1) {
        die("Error creating process.");
    } elseif ($pid == 0) {
        $generateIds();
        exit(0);
    } else {
        $childProcesses[] = $pid;
    }
}

// Wait for all processes to finish
foreach ($childProcesses as $pid) {
    pcntl_waitpid($pid, $status);
}

echo "Concurrency test finished.\n";

I have conducted 3 tests with different configurations in the Snowflake singleton.

Test 1

SequenceResolver: LaravelSequenceResolver
Cache driver: file
Singleton:

$this->app->singleton('snowflake', function (Application $app) {
    return (new Snowflake())
        ->setStartTimeStamp(strtotime('2019-10-10') * 1000)
        ->setSequenceResolver(new LaravelSequenceResolver($app->get('cache.store')));
});

Result: Many "Duplicate ID found: xxx" messages.

Test 2

SequenceResolver: LaravelSequenceResolver
Cache driver: redis
Singleton:

$this->app->singleton('snowflake', function (Application $app) {
    return (new Snowflake())
        ->setStartTimeStamp(strtotime('2019-10-10') * 1000)
        ->setSequenceResolver(new LaravelSequenceResolver($app->get('cache.store')));
});

Result: Many "Duplicate ID found: xxx" messages and some "RedisException read error on connection to redis:6379." messages.

Test 3

SequenceResolver: RedisSequenceResolver
Singleton:

$this->app->singleton('snowflake', function (Application $app) {
    $redis = new \Redis();
    $redis->connect(config('database.redis.cache.host'), (int) config('database.redis.cache.port'));
    $redis->auth(config('database.redis.cache.password'));
    $redis->select((int) config('database.redis.cache.database'));

    return (new Snowflake())
        ->setStartTimeStamp(strtotime('2019-10-10') * 1000)
        ->setSequenceResolver(new RedisSequenceResolver($redis));
});

Result: Works fine and no IDs are repeated.

Instead of a problem, I wanted to share my tests in case they are helpful to someone. In my case, Test 3 is the method that works for me and the one I'll be using. If anyone was able to successfully use the LaravelSequenceResolver as the SequenceResolver, I would like to know how you achieved it.

Greetings.

Currently, the key for cache entries in LaravelSequenceResolver and RedisSequenceResolver is just the current time as integer. If the Redis instance is share between machines it would mean that each of the machines would use the same cache entries leading to limitation of 256 unique ids per millisecond per all machines rather than 256 unique ids per machine.

What do you think about prefixing it with the machineId in case of Sonyflake and datacenter and workerId in case of Snowflake?

Hello,

there is a problem in the consistency of the defined return type.
Expected: Int, Given String.

PHPDoc gives the hint, that the return type will be of type int

    /**
     * Get snowflake id.
     *
     * @return int
     */
    public function id()
    {

(source)

Actual return is casted to string

        return (string) ((($currentTime - $this->getStartTimeStamp()) << $timestampLeftMoveLength)
            | ($this->datacenter << $datacenterLeftMoveLength)
            | ($this->workerid << $workerLeftMoveLength)
            | ($sequence));

(source)

for($i=1;$i<=10;$i++)
{
$snowflake = new \Godruoyi\Snowflake\Snowflake1, 1);
$snowflake->setStartTimeStamp($time);
$snowflake->setSequenceResolver(new RandomSequenceResolver());
return $snowflake->id();
}
生成的id都是一样的

Currently you can only get it as string, causing the id as an int to get transformed into a string, then back into an int again

Where is the Redis class, referenced in

首先感谢作者提供这么方便的包！

今天遇到了一个问题，后台生成的ID长度太大了，传递给前端之后，js因为数值太大失去了精度。请问修改什么配置能减小生成ID的长度

snowflake-7.lock is not a valid lock file. 报错是什么问题？

Why do I generate snowflake IDs the same when I put them in the for loop

2台机器，除了workId不一致，其他都是一样
比如都循环1万次，分别放置文件1.txt 和文件2.txt
单个文件sort 后没问题，还是1万条
但是合并后，在去重，显示19200+条，大概重复了400条，在线等，如何解决~~

I refer to the documentation ,but it comes to
Call to undefined method Illuminate\Support\Facades\Cache::store()
how to resolve it?

In the SnowFlake constructor the $datacenter and $workerId have a NULL default value.
So, if these values are not provided, the relevant properties are set to NULL ( the function mt_rand(0, 31) is not used ).
The NULL value for these properties seems to be a problem when the code tries to do bitwise shift . The result seems to be unstable.

For example when a execute:

$snowflake = new Snowflake();

for($i=1; $i<50; $i++) {
    $id = $snowflake->id();
    echo "$id \n";
    if ($i % 5 == 0) {
        sleep(2);
    }
}

the datacenter and worker part of the IDs is not the same for all iterations. 🤔
Am I doing something wrong ?

EDIT: My bad! I misunderstood something. Feel free to delete the issue.

I don't know how but it doesn't continue after setStartTimeStamp() function. Anyone know how to resolve?

$snowflake = new Godruoyi\Snowflake\Snowflake;
echo "seven";
try {
       $snowflake->setStartTimeStamp(strtotime('2019-09-09')*1000);
       echo $snowflake->getStartTimeStamp();
} catch (Exception $exception) {
       echo $exception;
}

echo "four";
$id = $snowflake->id();
echo $id;