dwood at softwarememetics.com
Sat Sep 16 16:54:47 UTC 2006
On 16 Sep2006, at 04:35, Leif Johansson wrote:
> On Saturdayen den 16 September 2006 04:40, Andrae Muys wrote:
>> On 16/09/2006, at 3:57 AM, Leif Johansson wrote:
>>>> This email is to encourage anyone working on Mulgara at the
>>>> moment to
>>>> discuss what you're doing in public. I'm also writing so that any
>>>> observers out there are aware that we're doing things, even if it
>>>> behind the scenes.
>>> I was going to postpone posting this until I had some more testing
>>> but this plea mustn't go unnoticed :-)
>>> I'm new to mulgara but have been keeping an eye on the project for
>>> a while
>>> since before it was mulgara. Anyways, I've written a J2EE resource
>>> for mulgara which essentially (at least it will when I have tested
>>> it) allows
>>> you to let a J2EE server manage connection pooling for you while
>>> a standardized way to reference a mulgara "connection" from a (say)
>>> an EJB.
>> That's cool. There were a number of discussions within Tucana before
>> it folded about how TKS could fit into J2EE. We started down that
>> road using Jotm and javax.transaction.* as the basis for our
>> transaction support internally. Of course we hadn't gone further
>> than this when Tucana closed. Thank you for this, I think it's
>> really great.
> At this stage the ra doesn't support transactions but it is next on my
> list - should be a snap if mulgaras transaction are ACID. The one
> I am unsure about is how the threading model in JCE will handle
> a local mulgara instance. Testing will tell.
Mulgara's transactions are *supposed* to be ACID, but there are two
current bugs. For the moment, you would have to go to Kowari's bug
tracker (at Sourceforge) and search for 'ACID' to find them:
These obviously need addressing in any case.
More information about the Mulgara-dev