[TransWarp] PROPOSAL: drop requirement for 'joinTxn=True' on SQL exec

alexander smishlajev alex at ank-sia.com
Fri Mar 7 10:45:17 EST 2003


Phillip J. Eby wrote:
> 
> So, I declare 'joinTxn=True' to be un-PEAKish, because:
> 
> 1) it requires you to be explicit about the common case, while being 
> implicit for a peculiar special case, that can actually be fixed using 
> pure SQL (using the "AT ISOLATION LEVEL" clause or a similar one in the 
> SQL itself),
> 
> 2) it has a "silent failure" mode, in that if you don't pay close 
> attention you may not notice anything is wrong,

ah yes.  i was hit by this too, a day before Ulrich Eck.  it hurts.

i have found the solution in mail archives thread starting from 
http://www.eby-sarna.com/pipermail/transwarp/2003-January/000276.html

i agree that 'joinTxn=True' requirement is bad.

> If there are no objections, then when I change the transaction-join 
> behavior I will also remove the "joinTxn" binding from the SQLCursor 
> class, so that using the 'joinTxn' flag will cause an error, thus 
> conveniently showing you where to remove this un-PEAKish wart from your 
> code.  :)

as for me, i found joinTxn feature very interesting: in many cases, i 
access data for read only, and i wonder if reading data without joining 
transaction may give some considerable performance improvement?

so i vote for leaving the "joinTxn" binding in the SQLCursor, but change 
default behaviour from 'joinTxn=False' to 'joinTxn=True'.  then PEAK 
users willing to disable transactional mechanisms will have an option to 
do so, but they must know what they're doing.

best wishes,
alex.





More information about the PEAK mailing list