Path: csiph.com!usenet.pasdenom.info!weretis.net!feeder4.news.weretis.net!eternal-september.org!feeder.eternal-september.org!mx04.eternal-september.org!.POSTED!not-for-mail
From: Eric Sosman <esosman@comcast-dot-net.invalid>
Newsgroups: comp.lang.java.programmer
Subject: Re: optimsed HashMap
Date: Tue, 27 Nov 2012 08:56:07 -0500
Organization: A noiseless patient Spider
Lines: 50
Message-ID: <k92gpu$2ah$1@dont-email.me>
References: <8i70b8d0pm6ibk03ti4t2pv60jd0bctlcs@4ax.com> <b9Tss.19809$W21.10542@newsfe27.iad> <k914ve$kl4$1@dont-email.me> <jc0ts.12172$v83.1077@newsfe09.iad>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Tue, 27 Nov 2012 13:56:14 +0000 (UTC)
Injection-Info: mx04.eternal-september.org; posting-host="ffb8f7085759b339c1002252b48331a4"; logging-data="2385"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/Do5dIeEosR5YNG0ptH1jH"
User-Agent: Mozilla/5.0 (Windows NT 5.1; rv:17.0) Gecko/17.0 Thunderbird/17.0
In-Reply-To: <jc0ts.12172$v83.1077@newsfe09.iad>
Cancel-Lock: sha1:sWqlpxaVL94uB6yVM6MswRHYWY4=
Xref: csiph.com comp.lang.java.programmer:19996

On 11/27/2012 5:01 AM, Arved Sandstrom wrote:
> On 11/26/2012 09:28 PM, Eric Sosman wrote:
>> On 11/26/2012 6:44 PM, Daniel Pitts wrote:
>>>
>>> I don't think your use-case needs this kind of (read "micro")
>>> optimization.  However, I have often wished for a way to get a Map.Entry
>>> for a key, which could then be used to insert a new value efficiently.
>>>
>>> Map.Entry<K,V> getOrAddEntry(K key);
>>>
>>> Map.Entry<K,V> e = map.getOrAddEntry(myKey);
>>> if (e.getValue() == null) {
>>>     e.setValue(newValue(myKey));
>>> }
>>> useValue(e.getValue());
>>>
>>> Alas, not exactly possible to add it now.
>>
>>      java.util.concurrent.ConcurrentMap has putIfAbsent().
>>
> It's almost certainly a reflection (no pun intended) of what I do with
> Maps, but a common variant use case for me is
>
> putEmptyListAndAddValueToListIfAbsent
>
> I don't normally use ConcurrentHashMap so the usual idiom for me is a
> contains() check that may put() the new empty list as a value,
> afterwards the list (empty or otherwise) is always got, and the actual
> value added to it.

     I usually write it as

	List<V> list = map.get(key);
	if (list == null) {
	    list = new SomeKindOfList();
	    map.put(key, list);
	}
	list.add(newValue);

... which involves one extra lookup per mapped key, not one
extra lookup per query.

     The imaginary getOrAddEntry() would be nice, although it
might complicate the Map.Entry interface itself (you'd want a
way to distinguish "This key maps to null" from "This key is
newly entered").

-- 
Eric Sosman
esosman@comcast-dot-net.invalid