[ Index ] |
PHP Cross Reference of MediaWiki-1.24.0 |
[Source view] [Print] [Project Stats]
This file deals with UID generation. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
Author: | Aaron Schulz |
File Size: | 507 lines (19 kb) |
Included or required: | 0 times |
Referenced: | 1 time |
Includes or requires: | 0 files |
UIDGenerator:: (18 methods):
__construct()
singleton()
newTimestampedUID88()
getTimestampedID88()
newTimestampedUID128()
getTimestampedID128()
newUUIDv4()
newRawUUIDv4()
newSequentialPerNodeID()
newSequentialPerNodeIDs()
getSequentialPerNodeIDs()
getTimestampAndDelay()
timeWaitUntil()
millisecondsSinceEpochBinary()
millitime()
deleteCacheFiles()
unitTestTearDown()
__destruct()
Class: UIDGenerator - X-Ref
Class for getting statistically unique IDs__construct() X-Ref |
No description |
singleton() X-Ref |
return: UIDGenerator |
newTimestampedUID88( $base = 10 ) X-Ref |
Get a statistically unique 88-bit unsigned integer ID string. The bits of the UID are prefixed with the time (down to the millisecond). These IDs are suitable as values for the shard key of distributed data. If a column uses these as values, it should be declared UNIQUE to handle collisions. New rows almost always have higher UIDs, which makes B-TREE updates on INSERT fast. They can also be stored "DECIMAL(27) UNSIGNED" or BINARY(11) in MySQL. UID generation is serialized on each server (as the node ID is for the whole machine). param: int $base Specifies a base other than 10 return: string Number |
getTimestampedID88( array $info ) X-Ref |
param: array $info (UIDGenerator::millitime(), counter, clock sequence) return: string 88 bits |
newTimestampedUID128( $base = 10 ) X-Ref |
Get a statistically unique 128-bit unsigned integer ID string. The bits of the UID are prefixed with the time (down to the millisecond). These IDs are suitable as globally unique IDs, without any enforced uniqueness. New rows almost always have higher UIDs, which makes B-TREE updates on INSERT fast. They can also be stored as "DECIMAL(39) UNSIGNED" or BINARY(16) in MySQL. UID generation is serialized on each server (as the node ID is for the whole machine). param: int $base Specifies a base other than 10 return: string Number |
getTimestampedID128( array $info ) X-Ref |
param: array $info (UIDGenerator::millitime(), counter, clock sequence) return: string 128 bits |
newUUIDv4( $flags = 0 ) X-Ref |
Return an RFC4122 compliant v4 UUID param: int $flags Bitfield (supports UIDGenerator::QUICK_RAND) return: string |
newRawUUIDv4( $flags = 0 ) X-Ref |
Return an RFC4122 compliant v4 UUID param: int $flags Bitfield (supports UIDGenerator::QUICK_RAND) return: string 32 hex characters with no hyphens |
newSequentialPerNodeID( $bucket, $bits = 48, $flags = 0 ) X-Ref |
Return an ID that is sequential *only* for this node and bucket These IDs are suitable for per-host sequence numbers, e.g. for some packet protocols. If UIDGenerator::QUICK_VOLATILE is used the counter might reset on server restart. param: string $bucket Arbitrary bucket name (should be ASCII) param: int $bits Bit size (<=48) of resulting numbers before wrap-around param: int $flags (supports UIDGenerator::QUICK_VOLATILE) return: float Integer value as float |
newSequentialPerNodeIDs( $bucket, $bits, $count, $flags = 0 ) X-Ref |
Return IDs that are sequential *only* for this node and bucket param: string $bucket Arbitrary bucket name (should be ASCII) param: int $bits Bit size (16 to 48) of resulting numbers before wrap-around param: int $count Number of IDs to return (1 to 10000) param: int $flags (supports UIDGenerator::QUICK_VOLATILE) return: array Ordered list of float integer values |
getSequentialPerNodeIDs( $bucket, $bits, $count, $flags ) X-Ref |
Return IDs that are sequential *only* for this node and bucket param: string $bucket Arbitrary bucket name (should be ASCII) param: int $bits Bit size (16 to 48) of resulting numbers before wrap-around param: int $count Number of IDs to return (1 to 10000) param: int $flags (supports UIDGenerator::QUICK_VOLATILE) return: array Ordered list of float integer values |
getTimestampAndDelay( $lockFile, $clockSeqSize, $counterSize ) X-Ref |
Get a (time,counter,clock sequence) where (time,counter) is higher than any previous (time,counter) value for the given clock sequence. This is useful for making UIDs sequential on a per-node bases. param: string $lockFile Name of a local lock file param: int $clockSeqSize The number of possible clock sequence values param: int $counterSize The number of possible counter values return: array (result of UIDGenerator::millitime(), counter, clock sequence) |
timeWaitUntil( array $time ) X-Ref |
Wait till the current timestamp reaches $time and return the current timestamp. This returns false if it would have to wait more than 10ms. param: array $time Result of UIDGenerator::millitime() return: array|bool UIDGenerator::millitime() result or false |
millisecondsSinceEpochBinary( array $time ) X-Ref |
param: array $time Result of UIDGenerator::millitime() return: string 46 MSBs of "milliseconds since epoch" in binary (rolls over in 4201) |
millitime() X-Ref |
return: array (current time in seconds, milliseconds since then) |
deleteCacheFiles() X-Ref |
Delete all cache files that have been created. This is a cleanup method primarily meant to be used from unit tests to avoid poluting the local filesystem. If used outside of a unit test environment it should be used with caution as it may destroy state saved in the files. |
unitTestTearDown() X-Ref |
Cleanup resources when tearing down after a unit test. This is a cleanup method primarily meant to be used from unit tests to avoid poluting the local filesystem. If used outside of a unit test environment it should be used with caution as it may destroy state saved in the files. |
__destruct() X-Ref |
No description |
Generated: Fri Nov 28 14:03:12 2014 | Cross-referenced by PHPXref 0.7.1 |