[PEAK] data manager, connections oh my

[Phillip J. Eby]

At 12:32 PM 1/7/04 -0600, darryl wrote:
>Hey,
>
>I have a bunch of stuff in a roundup tracker (http://roundup.sf.net) which 
>I want to access via roundup's api. My first thought was to
>write DataManagers which use the API to get my data in a PEAK-ish way.
>
>Now, i guess the problem is, that roundup sets a global lock on everything 
>when a connection is opened. What this means is while
>i have a connection open i cannot use roundup via the webUI.
>So what i need is a way to close the connection after every transaction.
>
>So i'm just throwing this out there for some guidance. Does the question 
>even make sense? Did i even ask a question?

Well, if you wrap your roundup connection in a ManagedConnection subclass, 
you can have it automatically close at the finish of each transaction.  All 
you should need to do are override in your subclass are _open(), _close(), 
cursorClass, and _closeASAP.  Something like:

class RoundupConnection(storage.ManagedConnection):

     _closeASAP = True   # force close after every transaction

     def _open(self):
          self.joinTxn() # ensure that we can only be opened in a transaction
          # now open the real roundup object and return it

     def _close(self):
         # do anything needed to 'self.connection' to close it

     cursorClass = RoundupCursor

Of course, you'll need to also create a RoundupCursor class, with an 
'execute()' method that takes positional arguments.  What it does with 
those arguments and how it returns results is up to you, of course.  You'll 
probably want to subclass AbstractCursor.  Note that keyword arguments 
passed to the connection's __call__ method are forwarded to the cursor 
class' constructor when creating cursors.