[tomjenkinson]

I am happy with that rationale, I more need confirmation that Paul is no longer wanting to remove those 66 calls (x2) and that we agree that removing "TransactionSynchronizationRegistryImple.getTransactionStatus" actually has zero effect so it is safe to remove.

In terms of BaseTransaction::getStatus, I am still on the fence. I think it should be logged for the scenario as explained but I just want to get consensus that is the only remaining item for discussion. As I mentioned above, most of the 600 calls to this actually take place outside the scope of the transaction that the tool is intending to debug. I would imagine the tool should be able to determine that and filter the unrelated ones out (they occur on different threads).

[paul.robinson]

Tom,

I'll need to have a think and do some investigation to confirm what you are suggesting. Leave it with me.

Paul.

[mmusgrov]

Tom Jenkinson wrote:

I am happy with that rationale, I more need confirmation that Paul is no longer wanting to remove those 66 calls (x2) and that we agree that removing "TransactionSynchronizationRegistryImple.getTransactionStatus" actually has zero effect so it is safe to remove.

In terms of BaseTransaction::getStatus, I am still on the fence. I think it should be logged for the scenario as explained but I just want to get consensus that is the only remaining item for discussion. As I mentioned above, most of the 600 calls to this actually take place outside the scope of the transaction that the tool is intending to debug. I would imagine the tool should be able to determine that and filter the unrelated ones out (they occur on different threads).

No, you misunderstood my point. I was arguing in response to you saying that some status values were of less interest than others. I argued "why is one status less important then another" and then I gave a rationale for why all status values are of potential interest, I was *not* saying that logging calls to getStatus is useful. I still stand by my earlier comment, quote:

I suspect the gets are not that useful (eg the state can be inferred since I believe we log each transaction state change) but the suspend probably is (unless like Tom says you can infer suspend from other log statements)

BTW I am not vetoing any proposed change, I am simply contributing my opinion that we are excessively noisy in our trace logs.

[tomjenkinson]

Hi Paul,

I removed TransactionSynchronizationRegistryImple.getTransactionStatus logging. Let me know what you want to do about the others. I really don't believe removing the logging for suspend/getTransactionKey will add anything as they are only logged comparatively infrequently. BaseTransaction.getStatus is still in discussion.

Thanks,

Tom