|
| 1 | +# Concurrent Ruby Notes |
| 2 | + |
| 3 | +## Locks |
| 4 | + |
| 5 | +Concurrent Ruby also has an internal extension of `Object` called |
| 6 | +`LockableObject`, which provides same synchronization primitives as Java's |
| 7 | +Object: `synchronize(&block)`, `wait(timeout = nil)`, |
| 8 | +`wait_until(timeout = nil, &condition)`, `signal`, `broadcast`. This class is |
| 9 | +intended for internal use in `concurrent-ruby` only and it does not support |
| 10 | +subclassing (since it cannot protect its lock from its children, for more |
| 11 | +details see [this article](http://wiki.apidesign.org/wiki/Java_Monitor)). It has |
| 12 | +minimal interface to be able to use directly locking available on given |
| 13 | +platforms. |
| 14 | + |
| 15 | +For non-internal use there is `Lock` and `Condition` implementation in |
| 16 | +`Synchronization` namespace, a condition can be obtained with `new_condition` |
| 17 | +method on `Lock`. So far their implementation is naive and requires more work. |
| 18 | +API is not expected to change. |
| 19 | + |
| 20 | +## Method names conventions |
| 21 | + |
| 22 | +Methods starting with `ns_` are marking methods that are not using |
| 23 | +synchronization by themselves, they have to be used inside synchronize block. |
| 24 | +They are usually used in pairs to separate the synchronization from behavior and |
| 25 | +to allow to call methods in the same object without double locking. |
| 26 | + |
| 27 | +``` ruby |
| 28 | +class Node |
| 29 | + # ... |
| 30 | + def left |
| 31 | + synchronize { ns_left } |
| 32 | + end |
| 33 | + |
| 34 | + def right |
| 35 | + synchronize { ns_right } |
| 36 | + end |
| 37 | + |
| 38 | + def to_a |
| 39 | + # avoids double locking |
| 40 | + synchronize { [ns_left, ns_right] } |
| 41 | + end |
| 42 | + |
| 43 | + private |
| 44 | + |
| 45 | + def ns_left |
| 46 | + @left |
| 47 | + end |
| 48 | + |
| 49 | + def ns_right |
| 50 | + @right |
| 51 | + end |
| 52 | + # ... |
| 53 | +end |
| 54 | +``` |
| 55 | +## Piggybacking |
| 56 | + |
| 57 | +Any write executed before volatile write based on program-order is visible to |
| 58 | +the volatile read as well, which allows |
| 59 | +[piggybacking](http://stackoverflow.com/questions/8769570/volatile-piggyback-is-this-enough-for-visiblity). |
| 60 | +Because it creates synchronizes-with (JMM term) order between volatile write |
| 61 | +and read, which participates in creating happens-before order. |
| 62 | + |
| 63 | +This trick is used in some of the abstractions, to avoid unnecessary |
| 64 | +synchronization or volatile declarations. |
0 commit comments