SQLObject

On Thu, Dec 01, 2005 at 01:11:44AM +0800, Yuan HOng wrote:
> It seems the select() method doesn't return a result set that reflects
> changes to the data rows made by other processes.

   Very much depends on transactions and SQLObject caching.

Oleg.
-- 
     Oleg Broytmann            http://phd.pp.ru/            phd@...
           Programmers don't die, they just GOSUB without RETURN.

On 11/30/05, Oleg Broytmann <phd@...> wrote:
> On Thu, Dec 01, 2005 at 01:11:44AM +0800, Yuan HOng wrote:
> > It seems the select() method doesn't return a result set that reflects
> > changes to the data rows made by other processes.
>
>    Very much depends on transactions and SQLObject caching.

I just wrote about this yesterday. If you enclose *all* of your
logical transactions (such as a web request) in a Transaction,
SQLObject's caching should not pose a problem for you, even in a
multiprocess environment.

Kevin
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Yuan HOng wrote:
> This means actually all column data is returned. However, the
> SQLObject.get method called by next() checks the id in cache, and if
> found, it discards the returned database data and returns the stale
> object in the cache.

If the stale object has changes waiting to be synced, wont' this
overwrite them?

-- 
Brian Beck
Adventurer of the First Order

Brian Beck wrote:
> Yuan HOng wrote:
> 
>>This means actually all column data is returned. However, the
>>SQLObject.get method called by next() checks the id in cache, and if
>>found, it discards the returned database data and returns the stale
>>object in the cache.
> 
> 
> If the stale object has changes waiting to be synced, wont' this
> overwrite them?

Your change, I mean.

-- 
Brian Beck
Adventurer of the First Order
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==

On Mon, Dec 12, 2005 at 12:55:38PM +0800, Yuan HOng wrote:
> I modified my patch so that only unchanged stales objects will be updated.
> 
> Index: main.py
> 
> ===================================================================
> 
> --- main.py     (revision 1386)
> 
> +++ main.py     (working copy)
> 
> @@ -960,6 +960,13 @@
> 
>                 cache.put(id, cls, val)
>             finally:
>                 cache.finishPut(cls)
> +        elif selectResults and not val.dirty:
> +            val._SO_writeLock.acquire()
> +            try:
> +                val._SO_selectInit(selectResults)
> +                val.sqlmeta.expired = False
> +            finally:
> +                val._SO_writeLock.release()
>         return val
> 
>     get = classmethod(get)

   Applied at the revision 1425 to the trunk, 1426 to the 0.7-branch. Thank
you.

Oleg.
-- 
     Oleg Broytmann            http://phd.pp.ru/            phd@...
           Programmers don't die, they just GOSUB without RETURN.