IRC log for #javaee, 2013-08-12

Please see http://irclog.greptilian.com/javaee for which days have been logged.

All times shown according to UTC.

Time	Nick	Message
00:47		SoniEx2\|2 joined ##javaee
03:46		SoniEx2\|2 joined ##javaee
04:46		SoniEx2\|2 joined ##javaee
06:42		Quest joined ##javaee
08:27		Quest joined ##javaee
10:05		sfisque joined ##javaee
12:46	Naros	morning sfisque
13:20		Naros joined ##javaee
13:21	Naros	morning all
13:29	pdurbin	mornin'
13:31	Naros	pdurbin: you have any experience with @EntityListeners?
13:34	pdurbin	nope
13:35	pdurbin	and `ack EntityListeners` returns nothing from our codebase
13:36	Naros	How about using things such as @PrePersist or @PostLoad or @PreUpdate, etc.
13:37	Naros	Essentially, @EntityListener and @EntityListeners allow you to define a callback class rather than muck your entities with JPA event callback methods.
13:37	pdurbin	hmm. ok. interesting
13:38	Naros	Anyway, looks like JPA instantiates these listeners on a per-entity basis.
13:38	Naros	which seems silly if they are suppose to be stateless.
13:39	Naros	I'd argue a cleaner approach would be to allocate one of each class and maintain some relationship between @Entity & @EntityListener
13:40	Naros	but can't find a way to make JPA do it :(
13:44	pdurbin	:(
13:46		SoniEx2 joined ##javaee
14:55	whartung	Naros: you can also put the entity listener methods in the actual entity, then this whole discussion is effectively moot.
14:56	whartung	the problem with making the listeners singletons, is then the framework writers are compelled to add Yet Another Lifecycle to start, stop, and tear down the Entity Listener, since the contract of a singleton is different from the contract of an instance variable.
14:57	whartung	if you want, you could always have a singleton service that the EntityListener leverages, but you'll still get an EL instance each time, however light weight it is.
14:58	whartung	But ELs tend to be little more than blobs of code, so the object over head is real small and, yea, they do add to GC pressure, marginally, but that's about as extreme a performance hit they add to the system.
15:03	pdurbin	the term EL is overloaded in Java EE :)
15:03	whartung	yes, but not in the context of this note.
15:14	Naros	whartung: understood. looks like in Hibernate 4.3, they instantiate each listener once by type where as in 4.2, they're instantiated by entity
15:14	whartung	ok
15:14	Naros	but let me ask you this
15:15	Naros	Lets say some entity has an embedded class
15:15		Quest joined ##javaee
15:15	whartung	biggest complain about ELs is that you can't actually do any JPA stuff in them...
15:15	whartung	you mean for the key?
15:15	Naros	No, not the key but actual repeatable column data.
15:15	whartung	ok
15:15	Quest	now we have two naros :)
15:16	Naros	Like a majority of entities perhaps have a set of 5 or 7 columns.
15:16	whartung	sure
15:16	Naros	I could allocate this embeddable in the private variable declaration of the entity.
15:16	Naros	but in the case of fetches, that is an unnecessary allocation since the ORM will do their own.
15:16	whartung	yes
15:16	Naros	so what I am trying to understand and follow are best practices for this use case.
15:17	Naros	Are those embeddables generally put off on the user of the persistence API to new when necessary during persists?
15:17	whartung	so you have
15:17	whartung	MyClass
15:17	whartung	which has
15:17	whartung	MyClass.get/setMyEmbedded
15:17	Naros	Right
15:17	whartung	right?
15:18	Naros	one way to avoid the unnecessary alloc is to have the get check if the field is null and if so, allocate and return the reference. that works for hibernate ofc, but maybe not another JPA provider.
15:18	whartung	so, then if you fetch a MyClass, the JPA should create a MyEmbedded as necessary, so myClass.getEmbedded will work
15:18	whartung	but if you do
15:18	whartung	myClass = new MyClass();
15:19	whartung	then, yes, it's up to you to create the MyEmbedded instance
15:19	whartung	and doing it in the getter is fine
15:19	whartung	if (myEmbedded == null) { myEmbedded = new MyEmbedded(); } return myEmbedded
15:19	Naros	Which do you believe is generally more acceptable, realizing there are many ways to cook this goose?
15:20	whartung	we use that patterns for collections
15:20	whartung	but I have never used embedded classes
15:20	Naros	the getter pattern?
15:20	whartung	JPA will NOT use your getter, since by default JPA is FIELD based. It'll never even see the getter
15:20	whartung	but the JPA will (should) create the instance on the fly for you anyway
15:21	Naros	Seems it does if annotations are on getters and not fields.
15:21	whartung	yea, you can change it
15:21	whartung	we don'
15:21	whartung	we just use the fields
15:21	whartung	and the get/setters for our code
15:21	Naros	Gotcha.
15:22	Naros	I was using some ELs to do some of this logic since they fire before the JPA provider's validation code fires but that feels overkill.
15:22	Naros	But wasn't sure if doing some null check in getters was taboo
15:22	whartung	shouldn't have to use the listeners at all for this
15:23	Naros	Aye, I feel the same. Listeners didn't feel right for this.
15:23	whartung	myClass = em.find(1); should Just Work like you think
15:23	Naros	Although some believe that is the perfect place to maintain date created and date last updated logic tho
15:24	whartung	and myClass.setEmbedded(new MyEmbedded()); em.persist(myClass) should just work as well
15:24	whartung	oh, sure listeners are fine for that
15:24	Naros	Aye, so long as I set it with a new, sure. but if I dont call the setter, it fails with error.
15:24	whartung	if the myEmbedded instance variable is null you get an error?
15:25	Naros	Aye, these two embeddables are AuditColumns which have the date craeted & last updated and DisabledColumns which handles entity disabled flags & disabled date logic.
15:25	whartung	in the MyClass?
15:25	whartung	ok
15:25	whartung	are they required?
15:25	Naros	Yes because for disabled flag the column declaration is
15:25	Naros	@Column(name="DISABLED_FLAG",nullable=false)
15:25	crimsonfubot	Naros: Error: "Column(name="DISABLED_FLAG",nullable=false)" is not a valid command.
15:25	Naros	grr
15:25	Naros	..@Column(name="DISABLED_FLAG",nullable=false)
15:25	Naros	yes, one is required in the column definition
15:26	whartung	so you have an embeddable with a required field within it
15:26	Naros	by default i wasn't it always to not be disabled and the user not have to worry with that
15:26	whartung	then have the instance be null should give you an error, no?
15:26	Naros	Yep
15:26	Naros	something about non-transient non-null value
15:26	whartung	huh
15:26	whartung	yea, you're getting in to uncharted territory for me here, since I've never used them
15:27	whartung	butI think you use case is a prefact use for these
15:27	Naros	Basically because the embeddable has a nullable=false definition on a column, the embeddable itself can't be null.
15:27	Naros	one i allocated the embeddable and set it, it persisted just fine.
15:27	Naros	*once
15:27	whartung	yea, semantically that makes sense, whether JPA groks that or relies on the DB to punt kittens when it happens, I can't say.
15:28	Naros	:P
15:28	Naros	i prefer punting gnomes myself
15:29	whartung	yea, not a fan of kitten abuse
15:30	whartung	here's your morning kitten productivity boost
15:30	whartung	https://i.chzbgr.com/maxW500/7726494976/h84FCA79A/
15:31	whartung	See...http://www.washingtonpost.com/blogs/wonkblog/wp/2012/10/01/want-to-increase-your-productivity-study-says-look-at-this-adorable-kitten/
15:31	whartung	SCIENCE! Don't be a denier...
15:32	Naros	ROFL -- I knew there was a reason I owned two adorable cats.
15:32	Naros	Never mind the fact that one weighs 17lbs and the other 8lbs.
15:33	whartung	yea we have two as well
15:36		sfisque joined ##javaee
15:47		Quest left ##javaee
15:52		kobain joined ##javaee
20:37		joshua_jandyco joined ##javaee
20:48		joshua_jandyco joined ##javaee
21:31		jxriddle joined ##javaee
21:33		jxriddle left ##javaee
21:46		SoniEx2 joined ##javaee
22:05		crimsonfubot joined ##javaee
22:05		kobain_ joined ##javaee
22:12		kobain joined ##javaee
22:29		kobain_ joined ##javaee
23:29		sfisque joined ##javaee
23:35		sfisque1 joined ##javaee

Please see http://irclog.greptilian.com/javaee for which days have been logged.