[cdi-dev] cyclic references

Mark Struberg struberg at yahoo.de
Mon Nov 26 09:18:40 EST 2012


yes, accessing via calling it ;)

LieGrue,
strub




----- Original Message -----
> From: Pete Muir <pmuir at redhat.com>
> To: Mark Struberg <struberg at yahoo.de>
> Cc: cdi-dev <cdi-dev at lists.jboss.org>
> Sent: Monday, November 26, 2012 2:53 PM
> Subject: Re: [cdi-dev] cyclic references
> 
> I don't believe the spec is updated for this. BTW I assume you mean 
> *calling* cyclic references from those methods?
> 
> On 24 Nov 2012, at 16:58, Mark Struberg wrote:
> 
>>  Hi Folks!
>> 
>>  I know we already discussed that cyclic references in @PostConstruct and 
> @Inject methods are not allowed (btw, did we also nail this down in the spec 
> properly? This is implicitly defined in commons-annotations).
>> 
>>  But what about @PreDestroy? As the destroy order is not defined for 
> normalscoped beans it could happen that a bean already got wiped and destroyed 
> while another one likes to access it. So it's not only a problem of cycles, 
> but might also happen without.
>> 
>>  sample:
>> 
>>  @RequestScoped
>> 
>>  public class BeanA {
>>    @Inject BeanB b;
>> 
>>    @PreDestroy
>>    public void destroyMe() {
>>      b.dosomething();
>>    }
>>  }
>> 
>>  and 
>> 
>> 
>>  @RequestScoped
>> 
>>  public class BeanB {
>>    public void dosomething() {..}
>> 
>>    @PreDestroy
>>    public void dropDatabaseConnection() {...}
>> 
>>  }
>> 
>>  Now lets imagine that BeanB got destroyed earlier than BeanA ....
>> 
>> 
>>  It's atm purely random. Imo this is ok, but should we point users to 
> this behaviour?
>> 
>>  LieGrue,
>>  strub
>> 
>> 
>>  _______________________________________________
>>  cdi-dev mailing list
>>  cdi-dev at lists.jboss.org
>>  https://lists.jboss.org/mailman/listinfo/cdi-dev
> 



More information about the cdi-dev mailing list