54 |
|
* <p> Resizing this or any other kind of hash table is a relatively |
55 |
|
* slow operation, so, when possible, it is a good idea to provide |
56 |
|
* estimates of expected table sizes in constructors. Also, for |
57 |
< |
* compatability with previous versions of this class, constructors |
57 |
> |
* compatibility with previous versions of this class, constructors |
58 |
|
* may optionally specify an expected {@code concurrencyLevel} as an |
59 |
|
* additional hint for internal sizing. |
60 |
|
* |
83 |
|
|
84 |
|
/** |
85 |
|
* A function computing a mapping from the given key to a value, |
86 |
< |
* or <code>null</code> if there is no mapping. This is a |
87 |
< |
* place-holder for an upcoming JDK8 interface. |
86 |
> |
* or {@code null} if there is no mapping. This is a place-holder |
87 |
> |
* for an upcoming JDK8 interface. |
88 |
|
*/ |
89 |
|
public static interface MappingFunction<K, V> { |
90 |
|
/** |
125 |
|
* within bins are always accurately traversable under volatile |
126 |
|
* reads, so long as lookups check hash code and non-nullness of |
127 |
|
* key and value before checking key equality. (All valid hash |
128 |
< |
* codes are nonnegative. Negative values are served for special |
129 |
< |
* nodes.) |
128 |
> |
* codes are nonnegative. Negative values are reserved for special |
129 |
> |
* forwarding nodes; see below.) |
130 |
|
* |
131 |
|
* A bin may be locked during update (insert, delete, and replace) |
132 |
|
* operations. We do not want to waste the space required to |
146 |
|
* there is no existing node during a put operation, then one can |
147 |
|
* be CAS'ed in (without need for lock except in computeIfAbsent); |
148 |
|
* the CAS serves as validation. This is on average the most |
149 |
< |
* common case for put operations. The expected number of locks |
150 |
< |
* covering different elements (i.e., bins with 2 or more nodes) |
151 |
< |
* is approximately 10% at steady state under default settings. |
152 |
< |
* Lock contention probability for two threads accessing arbitrary |
153 |
< |
* distinct elements is thus less than 1% even for small tables. |
149 |
> |
* common case for put operations -- under random hash codes, the |
150 |
> |
* distribution of nodes in bins follows a Poisson distribution |
151 |
> |
* (see http://en.wikipedia.org/wiki/Poisson_distribution) with a |
152 |
> |
* parameter of 0.5 on average under the default loadFactor of |
153 |
> |
* 0.75. The expected number of locks covering different elements |
154 |
> |
* (i.e., bins with 2 or more nodes) is approximately 10% at |
155 |
> |
* steady state under default settings. Lock contention |
156 |
> |
* probability for two threads accessing arbitrary distinct |
157 |
> |
* elements is, roughly, 1 / (8 * #elements). |
158 |
|
* |
159 |
|
* The table is resized when occupancy exceeds a threshold. Only |
160 |
|
* a single thread performs the resize (using field "resizing", to |
176 |
|
* complexity of access and iteration schemes that could admit |
177 |
|
* out-of-order or concurrent bin transfers. |
178 |
|
* |
179 |
< |
* (While not yet implemented, a similar traversal scheme can |
180 |
< |
* apply to partial traversals during partitioned aggregate |
181 |
< |
* operations. Also, read-only operations give up if ever |
182 |
< |
* forwarded to a null table, which provides support for |
183 |
< |
* shutdown-style clearing, which is also not currently |
180 |
< |
* implemented.) |
179 |
> |
* A similar traversal scheme (not yet implemented) can apply to |
180 |
> |
* partial traversals during partitioned aggregate operations. |
181 |
> |
* Also, read-only operations give up if ever forwarded to a null |
182 |
> |
* table, which provides support for shutdown-style clearing, |
183 |
> |
* which is also not currently implemented. |
184 |
|
* |
185 |
|
* The element count is maintained using a LongAdder, which avoids |
186 |
|
* contention on updates but can encounter cache thrashing if read |
187 |
|
* too frequently during concurrent updates. To avoid reading so |
188 |
< |
* often, resizing is attempted only upon adding to a bin already |
189 |
< |
* holding two or more nodes. Under the default threshold (0.75), |
190 |
< |
* and uniform hash distributions, the probability of this |
188 |
> |
* often, resizing is normally attempted only upon adding to a bin |
189 |
> |
* already holding two or more nodes. Under the default threshold |
190 |
> |
* (0.75), and uniform hash distributions, the probability of this |
191 |
|
* occurring at threshold is around 13%, meaning that only about 1 |
192 |
|
* in 8 puts check threshold (and after resizing, many fewer do |
193 |
< |
* so). To increase the probablity that a resize occurs soon |
194 |
< |
* enough, we offset the threshold (see THRESHOLD_OFFSET) by the |
195 |
< |
* expected number of puts between checks. This is currently set |
196 |
< |
* to 8, in accord with the default load factor. In practice, this |
197 |
< |
* is rarely overridden, and in any case is close enough to other |
198 |
< |
* plausible values not to waste dynamic probablity computation |
193 |
> |
* so). But this approximation has high variance for small table |
194 |
> |
* sizes, so we check on any collision for sizes <= 64. Further, |
195 |
> |
* to increase the probability that a resize occurs soon enough, we |
196 |
> |
* offset the threshold (see THRESHOLD_OFFSET) by the expected |
197 |
> |
* number of puts between checks. This is currently set to 8, in |
198 |
> |
* accord with the default load factor. In practice, this is |
199 |
> |
* rarely overridden, and in any case is close enough to other |
200 |
> |
* plausible values not to waste dynamic probability computation |
201 |
|
* for more precision. |
202 |
|
*/ |
203 |
|
|
234 |
|
static final int DEFAULT_CONCURRENCY_LEVEL = 16; |
235 |
|
|
236 |
|
/** |
237 |
< |
* The count value to offset thesholds to compensate for checking |
237 |
> |
* The count value to offset thresholds to compensate for checking |
238 |
|
* for resizing only when inserting into bins with two or more |
239 |
|
* elements. See above for explanation. |
240 |
|
*/ |
271 |
|
transient Set<Map.Entry<K,V>> entrySet; |
272 |
|
transient Collection<V> values; |
273 |
|
|
274 |
< |
/** For serialization compatability. Null unless serialized; see below */ |
274 |
> |
/** For serialization compatibility. Null unless serialized; see below */ |
275 |
|
Segment<K,V>[] segments; |
276 |
|
|
277 |
|
/** |
309 |
|
} |
310 |
|
|
311 |
|
/* |
312 |
< |
* Volatile access nethods are used for table elements as well as |
312 |
> |
* Volatile access methods are used for table elements as well as |
313 |
|
* elements of in-progress next table while resizing. Uses in |
314 |
|
* access and update methods are null checked by callers, and |
315 |
|
* implicitly bounds-checked, relying on the invariants that tab |
316 |
|
* arrays have non-zero size, and all indices are masked with |
317 |
|
* (tab.length - 1) which is never negative and always less than |
318 |
< |
* length. The only other usage is in HashIterator.advance, which |
319 |
< |
* performs explicit checks. |
318 |
> |
* length. The "relaxed" non-volatile forms are used only during |
319 |
> |
* table initialization. The only other usage is in |
320 |
> |
* HashIterator.advance, which performs explicit checks. |
321 |
|
*/ |
322 |
|
|
323 |
|
static final Node tabAt(Node[] tab, int i) { // used in HashIterator |
332 |
|
UNSAFE.putObjectVolatile(tab, ((long)i<<ASHIFT)+ABASE, v); |
333 |
|
} |
334 |
|
|
335 |
+ |
private static final Node relaxedTabAt(Node[] tab, int i) { |
336 |
+ |
return (Node)UNSAFE.getObject(tab, ((long)i<<ASHIFT)+ABASE); |
337 |
+ |
} |
338 |
+ |
|
339 |
+ |
private static final void relaxedSetTabAt(Node[] tab, int i, Node v) { |
340 |
+ |
UNSAFE.putObject(tab, ((long)i<<ASHIFT)+ABASE, v); |
341 |
+ |
} |
342 |
+ |
|
343 |
|
/* ---------------- Access and update operations -------------- */ |
344 |
|
|
345 |
< |
/** Implements get and containsKey **/ |
345 |
> |
/** Implementation for get and containsKey **/ |
346 |
|
private final Object internalGet(Object k) { |
347 |
|
int h = spread(k.hashCode()); |
348 |
|
Node[] tab = table; |
355 |
|
if (ev != null && ek != null && (k == ek || k.equals(ek))) |
356 |
|
return ev; |
357 |
|
} |
358 |
< |
if (eh < 0) { // bin was moved during resize |
358 |
> |
else if (eh < 0) { // bin was moved during resize |
359 |
|
tab = (Node[])e.key; |
360 |
|
continue retry; |
361 |
|
} |
366 |
|
return null; |
367 |
|
} |
368 |
|
|
369 |
< |
/** Implements put and putIfAbsent **/ |
369 |
> |
/** Implementation for put and putIfAbsent **/ |
370 |
|
private final Object internalPut(Object k, Object v, boolean replace) { |
371 |
|
int h = spread(k.hashCode()); |
372 |
|
Object oldVal = null; // the previous value or null if none |
359 |
– |
Node node = null; // the node created if absent |
373 |
|
Node[] tab = table; |
374 |
|
for (;;) { |
375 |
|
Node e; int i; |
376 |
|
if (tab == null) |
377 |
|
tab = grow(0); |
378 |
|
else if ((e = tabAt(tab, i = (tab.length - 1) & h)) == null) { |
379 |
< |
if (node == null) |
367 |
< |
node = new Node(h, k, v, null); |
368 |
< |
if (casTabAt(tab, i, null, node)) |
379 |
> |
if (casTabAt(tab, i, null, new Node(h, k, v, null))) |
380 |
|
break; |
381 |
|
} |
382 |
|
else if (e.hash < 0) |
384 |
|
else { |
385 |
|
boolean validated = false; |
386 |
|
boolean checkSize = false; |
387 |
< |
synchronized(e) { |
387 |
> |
synchronized (e) { |
388 |
|
Node first = e; |
389 |
|
for (;;) { |
390 |
|
Object ek, ev; |
404 |
|
if ((e = e.next) == null) { |
405 |
|
if (tabAt(tab, i) == first) { |
406 |
|
validated = true; |
407 |
< |
if (node == null) |
408 |
< |
node = new Node(h, k, v, null); |
398 |
< |
last.next = node; |
399 |
< |
if (last != first) |
407 |
> |
last.next = new Node(h, k, v, null); |
408 |
> |
if (last != first || tab.length <= 64) |
409 |
|
checkSize = true; |
410 |
|
} |
411 |
|
break; |
442 |
|
else { |
443 |
|
boolean validated = false; |
444 |
|
boolean deleted = false; |
445 |
< |
synchronized(e) { |
445 |
> |
synchronized (e) { |
446 |
|
Node pred = null; |
447 |
|
Node first = e; |
448 |
|
for (;;) { |
485 |
|
return oldVal; |
486 |
|
} |
487 |
|
|
488 |
< |
/** Implements computeIfAbsent */ |
488 |
> |
/** Implementation for computeIfAbsent and compute */ |
489 |
|
@SuppressWarnings("unchecked") |
490 |
< |
private final V computeVal(K k, MappingFunction<? super K, ? extends V> f) { |
490 |
> |
private final V internalCompute(K k, |
491 |
> |
MappingFunction<? super K, ? extends V> f, |
492 |
> |
boolean replace) { |
493 |
|
int h = spread(k.hashCode()); |
494 |
|
V val = null; |
484 |
– |
Node node = null; |
495 |
|
boolean added = false; |
496 |
|
boolean validated = false; |
497 |
|
Node[] tab = table; |
500 |
|
if (tab == null) |
501 |
|
tab = grow(0); |
502 |
|
else if ((e = tabAt(tab, i = (tab.length - 1) & h)) == null) { |
503 |
< |
if (node == null) |
504 |
< |
node = new Node(h, k, null, null); |
495 |
< |
synchronized(node) { |
503 |
> |
Node node = new Node(h, k, null, null); |
504 |
> |
synchronized (node) { |
505 |
|
if (casTabAt(tab, i, null, node)) { |
506 |
|
validated = true; |
507 |
|
try { |
519 |
|
} |
520 |
|
else if (e.hash < 0) |
521 |
|
tab = (Node[])e.key; |
522 |
+ |
else if (Thread.holdsLock(e)) |
523 |
+ |
throw new IllegalStateException("Recursive map computation"); |
524 |
|
else { |
525 |
|
boolean checkSize = false; |
526 |
< |
synchronized(e) { |
526 |
> |
synchronized (e) { |
527 |
|
Node first = e; |
528 |
|
for (;;) { |
529 |
|
Object ek, ev; |
533 |
|
(k == ek || k.equals(ek))) { |
534 |
|
if (tabAt(tab, i) == first) { |
535 |
|
validated = true; |
536 |
+ |
if (replace && (ev = f.map(k)) != null) |
537 |
+ |
e.val = ev; |
538 |
|
val = (V)ev; |
539 |
|
} |
540 |
|
break; |
544 |
|
if (tabAt(tab, i) == first) { |
545 |
|
validated = true; |
546 |
|
if ((val = f.map(k)) != null) { |
547 |
< |
if (node == null) |
535 |
< |
node = new Node(h, k, val, null); |
536 |
< |
else |
537 |
< |
node.val = val; |
538 |
< |
last.next = node; |
539 |
< |
if (last != first) |
540 |
< |
checkSize = true; |
547 |
> |
last.next = new Node(h, k, val, null); |
548 |
|
added = true; |
549 |
+ |
if (last != first || tab.length <= 64) |
550 |
+ |
checkSize = true; |
551 |
|
} |
552 |
|
} |
553 |
|
break; |
592 |
|
} |
593 |
|
else { |
594 |
|
boolean validated = false; |
595 |
< |
synchronized(e) { |
595 |
> |
synchronized (e) { |
596 |
|
int idx = e.hash & mask; |
597 |
|
Node lastRun = e; |
598 |
|
for (Node p = e.next; p != null; p = p.next) { |
604 |
|
} |
605 |
|
if (tabAt(tab, i) == e) { |
606 |
|
validated = true; |
607 |
< |
setTabAt(nextTab, idx, lastRun); |
607 |
> |
relaxedSetTabAt(nextTab, idx, lastRun); |
608 |
|
for (Node p = e; p != lastRun; p = p.next) { |
609 |
|
int h = p.hash; |
610 |
|
int j = h & mask; |
611 |
< |
Object pk = p.key, pv = p.val; |
612 |
< |
Node r = tabAt(nextTab, j); |
613 |
< |
setTabAt(nextTab, j, new Node(h, pk, pv, r)); |
611 |
> |
Node r = relaxedTabAt(nextTab, j); |
612 |
> |
relaxedSetTabAt(nextTab, j, |
613 |
> |
new Node(h, p.key, p.val, r)); |
614 |
|
} |
615 |
|
setTabAt(tab, i, fwd); |
616 |
|
} |
632 |
|
* @return current table |
633 |
|
*/ |
634 |
|
private final Node[] grow(int sizeHint) { |
626 |
– |
Node[] tab; |
635 |
|
if (resizing == 0 && |
636 |
|
UNSAFE.compareAndSwapInt(this, resizingOffset, 0, 1)) { |
637 |
|
try { |
638 |
|
for (;;) { |
639 |
|
int cap, n; |
640 |
< |
if ((tab = table) == null) { |
640 |
> |
Node[] tab = table; |
641 |
> |
if (tab == null) { |
642 |
|
int c = initCap; |
643 |
|
if (c < sizeHint) |
644 |
|
c = sizeHint; |
662 |
|
if (tab != null) |
663 |
|
transfer(tab, nextTab); |
664 |
|
table = nextTab; |
665 |
< |
if (tab == null || counter.sum() < threshold) { |
666 |
< |
tab = nextTab; |
665 |
> |
if (tab == null || cap >= MAXIMUM_CAPACITY || |
666 |
> |
(sizeHint > 0 && cap >= sizeHint) || |
667 |
> |
counter.sum() < threshold) |
668 |
|
break; |
659 |
– |
} |
669 |
|
} |
670 |
|
} finally { |
671 |
|
resizing = 0; |
672 |
|
} |
673 |
|
} |
674 |
< |
else if ((tab = table) == null) |
674 |
> |
else if (table == null) |
675 |
|
Thread.yield(); // lost initialization race; just spin |
676 |
< |
return tab; |
676 |
> |
return table; |
677 |
|
} |
678 |
|
|
679 |
|
/** |
680 |
< |
* Implements putAll and constructor with Map argument. Tries to |
681 |
< |
* first override initial capacity or grow (once) based on map |
682 |
< |
* size to pre-allocate table space. |
680 |
> |
* Implementation for putAll and constructor with Map |
681 |
> |
* argument. Tries to first override initial capacity or grow |
682 |
> |
* based on map size to pre-allocate table space. |
683 |
|
*/ |
684 |
|
private final void internalPutAll(Map<? extends K, ? extends V> m) { |
685 |
|
int s = m.size(); |
686 |
< |
grow((s >= (MAXIMUM_CAPACITY >>> 1))? s : s + (s >>> 1)); |
686 |
> |
grow((s >= (MAXIMUM_CAPACITY >>> 1)) ? s : s + (s >>> 1)); |
687 |
|
for (Map.Entry<? extends K, ? extends V> e : m.entrySet()) { |
688 |
|
Object k = e.getKey(); |
689 |
|
Object v = e.getValue(); |
694 |
|
} |
695 |
|
|
696 |
|
/** |
697 |
< |
* Implements clear. Steps through each bin, removing all nodes. |
697 |
> |
* Implementation for clear. Steps through each bin, removing all nodes. |
698 |
|
*/ |
699 |
|
private final void internalClear() { |
700 |
< |
long delta = 0L; // negative of number of deletions |
700 |
> |
long deletions = 0L; |
701 |
|
int i = 0; |
702 |
|
Node[] tab = table; |
703 |
|
while (tab != null && i < tab.length) { |
708 |
|
tab = (Node[])e.key; |
709 |
|
else { |
710 |
|
boolean validated = false; |
711 |
< |
synchronized(e) { |
711 |
> |
synchronized (e) { |
712 |
|
if (tabAt(tab, i) == e) { |
713 |
|
validated = true; |
714 |
|
do { |
715 |
|
if (e.val != null) { |
716 |
|
e.val = null; |
717 |
< |
--delta; |
717 |
> |
++deletions; |
718 |
|
} |
719 |
|
} while ((e = e.next) != null); |
720 |
|
setTabAt(tab, i, null); |
721 |
|
} |
722 |
|
} |
723 |
< |
if (validated) |
723 |
> |
if (validated) { |
724 |
|
++i; |
725 |
+ |
if (deletions > THRESHOLD_OFFSET) { // bound lag in counts |
726 |
+ |
counter.add(-deletions); |
727 |
+ |
deletions = 0L; |
728 |
+ |
} |
729 |
+ |
} |
730 |
|
} |
731 |
|
} |
732 |
< |
counter.add(delta); |
732 |
> |
if (deletions != 0L) |
733 |
> |
counter.add(-deletions); |
734 |
|
} |
735 |
|
|
736 |
|
/** |
977 |
|
* @return {@code true} if this map contains no key-value mappings |
978 |
|
*/ |
979 |
|
public boolean isEmpty() { |
980 |
< |
return counter.sum() == 0L; |
980 |
> |
return counter.sum() <= 0L; // ignore transient negative values |
981 |
|
} |
982 |
|
|
983 |
|
/** |
989 |
|
*/ |
990 |
|
public int size() { |
991 |
|
long n = counter.sum(); |
992 |
< |
return n >= Integer.MAX_VALUE ? Integer.MAX_VALUE : (int)n; |
992 |
> |
return ((n >>> 31) == 0) ? (int)n : (n < 0L) ? 0 : Integer.MAX_VALUE; |
993 |
|
} |
994 |
|
|
995 |
|
/** |
1118 |
|
* return map.get(key); |
1119 |
|
* value = mappingFunction.map(key); |
1120 |
|
* if (value != null) |
1121 |
< |
* return map.put(key, value); |
1122 |
< |
* else |
1108 |
< |
* return null; |
1121 |
> |
* map.put(key, value); |
1122 |
> |
* return value; |
1123 |
|
* </pre> |
1124 |
|
* |
1125 |
|
* except that the action is performed atomically. Some attempted |
1126 |
< |
* operations on this map by other threads may be blocked while |
1127 |
< |
* computation is in progress. Because this function is invoked |
1128 |
< |
* within atomicity control, the computation should be short and |
1129 |
< |
* simple, and must not attempt to update any other mappings of |
1130 |
< |
* this Map. The most common usage is to construct a new object |
1131 |
< |
* serving as an initial mapped value, or memoized result. |
1126 |
> |
* update operations on this map by other threads may be blocked |
1127 |
> |
* while computation is in progress, so the computation should be |
1128 |
> |
* short and simple, and must not attempt to update any other |
1129 |
> |
* mappings of this Map. The most appropriate usage is to |
1130 |
> |
* construct a new object serving as an initial mapped value, or |
1131 |
> |
* memoized result, as in: |
1132 |
> |
* <pre>{@code |
1133 |
> |
* map.computeIfAbsent(key, new MappingFunction<K, V>() { |
1134 |
> |
* public V map(K k) { return new Value(f(k)); }}; |
1135 |
> |
* }</pre> |
1136 |
|
* |
1137 |
|
* @param key key with which the specified value is to be associated |
1138 |
|
* @param mappingFunction the function to compute a value |
1141 |
|
* returned {@code null}. |
1142 |
|
* @throws NullPointerException if the specified key or mappingFunction |
1143 |
|
* is null, |
1144 |
+ |
* @throws IllegalStateException if the computation detectably |
1145 |
+ |
* attempts a recursive update to this map that would |
1146 |
+ |
* otherwise never complete. |
1147 |
|
* @throws RuntimeException or Error if the mappingFunction does so, |
1148 |
|
* in which case the mapping is left unestablished. |
1149 |
|
*/ |
1150 |
|
public V computeIfAbsent(K key, MappingFunction<? super K, ? extends V> mappingFunction) { |
1151 |
|
if (key == null || mappingFunction == null) |
1152 |
|
throw new NullPointerException(); |
1153 |
< |
return computeVal(key, mappingFunction); |
1153 |
> |
return internalCompute(key, mappingFunction, false); |
1154 |
> |
} |
1155 |
> |
|
1156 |
> |
/** |
1157 |
> |
* Computes the value associated with the given key using the given |
1158 |
> |
* mappingFunction, and if non-null, enters it into the map. This |
1159 |
> |
* is equivalent to |
1160 |
> |
* |
1161 |
> |
* <pre> |
1162 |
> |
* value = mappingFunction.map(key); |
1163 |
> |
* if (value != null) |
1164 |
> |
* map.put(key, value); |
1165 |
> |
* else |
1166 |
> |
* value = map.get(key); |
1167 |
> |
* return value; |
1168 |
> |
* </pre> |
1169 |
> |
* |
1170 |
> |
* except that the action is performed atomically. Some attempted |
1171 |
> |
* update operations on this map by other threads may be blocked |
1172 |
> |
* while computation is in progress, so the computation should be |
1173 |
> |
* short and simple, and must not attempt to update any other |
1174 |
> |
* mappings of this Map. |
1175 |
> |
* |
1176 |
> |
* @param key key with which the specified value is to be associated |
1177 |
> |
* @param mappingFunction the function to compute a value |
1178 |
> |
* @return the current value associated with |
1179 |
> |
* the specified key, or {@code null} if the computation |
1180 |
> |
* returned {@code null} and the value was not otherwise present. |
1181 |
> |
* @throws NullPointerException if the specified key or mappingFunction |
1182 |
> |
* is null, |
1183 |
> |
* @throws IllegalStateException if the computation detectably |
1184 |
> |
* attempts a recursive update to this map that would |
1185 |
> |
* otherwise never complete. |
1186 |
> |
* @throws RuntimeException or Error if the mappingFunction does so, |
1187 |
> |
* in which case the mapping is unchanged. |
1188 |
> |
*/ |
1189 |
> |
public V compute(K key, MappingFunction<? super K, ? extends V> mappingFunction) { |
1190 |
> |
if (key == null || mappingFunction == null) |
1191 |
> |
throw new NullPointerException(); |
1192 |
> |
return internalCompute(key, mappingFunction, true); |
1193 |
|
} |
1194 |
|
|
1195 |
|
/** |
1205 |
|
public V remove(Object key) { |
1206 |
|
if (key == null) |
1207 |
|
throw new NullPointerException(); |
1208 |
< |
return (V)internalReplace(key, null, null); |
1208 |
> |
return (V)internalReplace(key, null, null); |
1209 |
|
} |
1210 |
|
|
1211 |
|
/** |
1229 |
|
public boolean replace(K key, V oldValue, V newValue) { |
1230 |
|
if (key == null || oldValue == null || newValue == null) |
1231 |
|
throw new NullPointerException(); |
1232 |
< |
return internalReplace(key, newValue, oldValue) != null; |
1232 |
> |
return internalReplace(key, newValue, oldValue) != null; |
1233 |
|
} |
1234 |
|
|
1235 |
|
/** |
1243 |
|
public V replace(K key, V value) { |
1244 |
|
if (key == null || value == null) |
1245 |
|
throw new NullPointerException(); |
1246 |
< |
return (V)internalReplace(key, value, null); |
1246 |
> |
return (V)internalReplace(key, value, null); |
1247 |
|
} |
1248 |
|
|
1249 |
|
/** |
1337 |
|
} |
1338 |
|
|
1339 |
|
/** |
1340 |
< |
* {@inheritDoc} |
1340 |
> |
* Returns the hash code value for this {@link Map}, i.e., |
1341 |
> |
* the sum of, for each key-value pair in the map, |
1342 |
> |
* {@code key.hashCode() ^ value.hashCode()}. |
1343 |
> |
* |
1344 |
> |
* @return the hash code value for this map |
1345 |
|
*/ |
1346 |
|
public int hashCode() { |
1347 |
|
return new HashIterator().mapHashCode(); |
1348 |
|
} |
1349 |
|
|
1350 |
|
/** |
1351 |
< |
* {@inheritDoc} |
1351 |
> |
* Returns a string representation of this map. The string |
1352 |
> |
* representation consists of a list of key-value mappings (in no |
1353 |
> |
* particular order) enclosed in braces ("{@code {}}"). Adjacent |
1354 |
> |
* mappings are separated by the characters {@code ", "} (comma |
1355 |
> |
* and space). Each key-value mapping is rendered as the key |
1356 |
> |
* followed by an equals sign ("{@code =}") followed by the |
1357 |
> |
* associated value. |
1358 |
> |
* |
1359 |
> |
* @return a string representation of this map |
1360 |
|
*/ |
1361 |
|
public String toString() { |
1362 |
|
return new HashIterator().mapToString(); |
1363 |
|
} |
1364 |
|
|
1365 |
|
/** |
1366 |
< |
* {@inheritDoc} |
1366 |
> |
* Compares the specified object with this map for equality. |
1367 |
> |
* Returns {@code true} if the given object is a map with the same |
1368 |
> |
* mappings as this map. This operation may return misleading |
1369 |
> |
* results if either map is concurrently modified during execution |
1370 |
> |
* of this method. |
1371 |
> |
* |
1372 |
> |
* @param o object to be compared for equality with this map |
1373 |
> |
* @return {@code true} if the specified object is equal to this map |
1374 |
|
*/ |
1375 |
|
public boolean equals(Object o) { |
1376 |
|
if (o == this) |