Re: Proposal for JDO-259

[Craig L Russell]

Hi Michael, On Jan 16, 2006, at 7:17 AM, Michael Watzek wrote: Hi, JDO-259 contains 4 assertions related to state persistent-nontransactional-dirty. Below, you find a proposal on how these assertions may be implemented. Please note, that these assertions do not specify life cycle transitions. Instead, they specify flushing of persistent-nontransactional-dirty instances into the datastore at commit/rollback. For this reason, I propose to implement new test cases instead of incorporating the test cases into the existing life cycle tests. Right. These are behavior assertions that should be distinct from lifecycle assertions and should be in a separate test from the lifecycle tests. ----------- A5.6.2-4 [If a datastore transaction is begun, commit will write the changes to the datastore with no checking as to the current state of the instances in the datastore. That is, the changes made outside the transaction together with any changes made inside the transaction will overwrite the current state of the datastore.] Test case: // initially make an instance persistent pm1.begin pm1.makePersistent(pc) oid = pm1.getObjectId(pc) pm1.commit // ensure concurrent changes of the instance in the datastore pm2.begin pc2 = pm2.getObjectById(oid) modify(pc2) pm2.commit // make the instance persistent-nontransactional-dirty Just for clarity, perhaps this modify block should be moved before the pm2 changes. Then it's clear that the state being modified is not in any way related to the changes in pm2. modify(pc) // flush the instance into the datastore pm1.begin pm1.commit // check the state in the datastore It should match the changes made by pm1 not the changes made by pm2. check(pm3.getObjectById(oid)) ----------- A5.6.2-6 [If a datastore transaction is begun, rollback will not write any changes to the datastore.] Test case: // initially make an instance persistent pm1.begin pm1.makePersistent(pc) oid = pm1.getObjectId(pc) pm1.commit // make the instance persistent-nontransactional-dirty modify(pc) // discard the changes pm1.begin pm1.rollback // check the state in the datastore check(pm3.getObjectById(oid)) ----------- A5.6.2-8 [If an optimistic transaction is begun, commit will write the changes to the datastore after checking as to the current state of the instances in the datastore. The changes made outside the transaction together with any changes made inside the transaction will update the current state of the datastore if the version checking is successful.] Test case w/o concurrent changes: // initially make an instance persistent pm1.begin pm1.makePersistent(pc) oid = pm1.getObjectId(pc) pm1.commit // make the instance persistent-nontransactional-dirty modify(pc) // flush the instance into the datastore pm1.begin(optimistic==true) pm1.commit // check the state in the datastore check(pm3.getObjectById(oid)) Test case w/ concurrent changes: // initially make an instance persistent pm1.begin pm1.makePersistent(pc) oid = pm1.getObjectId(pc) pm1.commit // ensure concurrent changes of the instance in the datastore pm2.begin pc2 = pm2.getObjectById(oid) modify(pc2) pm2.commit // make the instance persistent-nontransactional-dirty modify(pc) // flush the instance into the datastore pm1.begin(optimistic==true) pm1.commit // check the state in the datastore check(pm3.getObjectById(oid)) ----------- A5.6.2-10 [If an optimistic transaction is begun, rollback will not write any changes to the datastore. The persistent-nontransactional-dirty instances will transition according to the RestoreValues flag.] The state transition should be a different assertion. See below. Test case: // initially make an instance persistent pm1.begin pm1.makePersistent(pc) oid = pm1.getObjectId(pc) pm1.commit // make the instance persistent-nontransactional-dirty modify(pc) // discard the changes pm1.begin(optimistic==true) pm1.rollback // check the state in the datastore check(pm3.getObjectById(oid)) ----------- Questions: 1) The last assertion (A5.6.2-10) contains the sentence: "The persistent-nontransactional-dirty instances will transition according to the RestoreValues flag." I propose to drop the sentence from the assertion because the semantics of the RestoreValues flag is checked by other assertions related to JDO-258. I agree. The spec assertion is wrong. I've changed the spec as follows: A5.6.2-10 [If an optimistic transaction is begun, rollback will not write any changes to the datastore. ] A5.6.2-11 [The persistent-nontransactional-dirty instances will transition according to the RestoreValues flag. ] A5.6.2-12 [With the RestoreValues flag set to true, persistent-nontransactional-dirty instances will make no state transition, ] A5.6.2-13 [but the fields will be restored to their values as of the beginning of the transaction, and any changes made within the transaction will be discarded. ] A5.6.2-14 [With the RestoreValues flag set to false, persistent-nontransactional-dirty instances will transition to hollow.]  Craig Regards, Michael --  ------------------------------------------------------------------- Michael Watzek                  [EMAIL PROTECTED] Engineering GmbH mailto:[EMAIL PROTECTED]        Buelowstr. 66 Tel.:  ++49/30/235 520 36       10783 Berlin - Germany Fax.:  ++49/30/217 520 12       http://www.spree.de/ ------------------------------------------------------------------- Craig Russell Architect, Sun Java Enterprise System http://java.sun.com/products/jdo 408 276-5638 mailto:[EMAIL PROTECTED] P.S. A good JDO? O, Gasp!

smime.p7s
Description: S/MIME cryptographic signature