No, startT makes no claim (and most likely can't make such a claim) to clean
up properly if it encounters an exception.
On 11/12/07 10:57 AM, "Toby Thain" <[EMAIL PROTECTED]> wrote:
>
> On 12-Nov-07, at 3:56 PM, Toby Thain wrote:
>
>>
>> On 12-Nov-07, at 3:27 PM, Gwyn Evans wrote:
>>
>>> No idea if it's possible, but the first case allows for the
>>> transaction being created then an exception being thrown in the
>>> 'start' operation, whereas with the second case, that would leave the
>>> transaction open.
>>
>> Wouldn't that be a bug?
>
> ...in startTransaction(), I mean...
>
> --Toby
>
>>
>> --Toby
>>
>>>
>>> /Gwyn
>>>
>>> On 12/11/2007, Toby Thain <[EMAIL PROTECTED]> wrote:
>>>> Hi,
>>>>
>>>> A little thought has been bothering me for a while and I thought I
>>>> would put it to the experts.
>>>>
>>>> The iBATIS manual advises the following logic for transactions:
>>>>
>>>> try {
>>>> sqlMap.startTransaction ();
>>>> // ...queries
>>>> sqlMap.commitTransaction ();
>>>> } finally {
>>>> sqlMap.endTransaction ();
>>>> }
>>>>
>>>> I am curious if it would be more strictly correct to:
>>>>
>>>> sqlMap.startTransaction ();
>>>> try {
>>>> // ...queries
>>>> sqlMap.commitTransaction ();
>>>> } finally {
>>>> sqlMap.endTransaction ();
>>>> }
>>>>
>>>> In the event that startTransaction() itself throws an Exception...
>>>>
>>>> Thoughts?
>>>> --Toby
>>>>
>>>
>>>
>>> --
>>> Download Wicket 1.3.0-rc1 now! - http://wicketframework.org
>>
>