Well, because table 6 clearly states "Write instance state back to database
= Yes" for each commit option. It does indeed sound strange to me, and is
not in synch with my experience of some containers. But looking solely at
the specification, I can't help getting the impression that it would indeed
require writing bean contents to database even when state has not been
changed.
Personally I'm leaning towards the conclusion that the specification is
written a bit badly in that place. Of course, that's subjective.
johan
<[EMAIL PROTECTED] To: [EMAIL PROTECTED]
ista.se> cc: [EMAIL PROTECTED]
Subject: Re: CMP and writing instance
back to database
17.04.2004 16:15
Why would you assume that "synchronizes the instance's state with the
persistent storage" should require the container to force beans to
persistent store even though the container knows the state is already
in sync (because it was never altered)?
/johan
On 2004-04-17, at 10.20, Janne Mattila wrote:
> Some confusion about the EJB specification:
>
> EJB 2.1 specification, 10.5.10, "commit options" states that "container
> synchronizes the instance's state with the persistent storage at
> transaction commit for all three (commit) options". Table 6 states this
> even more explicitly, having "Yes" at "Write instance state to
> database"
> for all commit options. Reading this, I would assume that the
> specification
> requires CMP engine to write each bean back to database at commit
> time, no
> matter what.
>
> This in contradiction with my experiences with WebLogic. WebLogic 6.1
> "...Server CMP implementation automatically detects modifications of
> CMP
> fields and writes only those changes to the underlying datastore.".
> This
> seems logical, as writing CMP fields back when they have not been
> changed
> would not be efficient.
>
> Does anyone have an idea on what is really going on here:
>
> a) the intent of specification indeed requires that all fields be
> written
> back to db at commit, but WebLogic has chosen not to follow the
> specification to ensure better performance
> b) specification is written a bit badly, and in fact it requires that
> ejbStore() be called at commit time - and WebLogic follows that
> specification properly
> c) something else?
>
> =======================================================================
> ====
> To unsubscribe, send email to [EMAIL PROTECTED] and include in the
> body
> of the message "signoff EJB-INTEREST". For general help, send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
>
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
