In other transactional systems I've seen, the framework provides decorators you can apply to your controller methods, specifying the level of transaction isolation you need. These 'transaction isolation levels' are alternative means to specifying 'for update' on your SQL query and affect *every* query in the transaction. For more info, see here: http://en.wikipedia.org/wiki/Isolation_(computer_science).
Does SqlAlchemy (or SqlObject, for that matter) provide such decorators? If not, is there some other way to set this on a per-method basis? Barry ----- Original Message ---- From: tml <[EMAIL PROTECTED]> To: TurboGears <[email protected]> Sent: Thursday, March 15, 2007 9:45:05 AM Subject: [TurboGears] Re: incrementing in visitor counter Hello guys, I tried what Andy suggested and its not working. Basically had two shells: In [1]: u = User.get(1) In [2]: u Out[2]: <tgWeb.model.User object at 0xb725f40c> In [3]: u.visit_counter += 1 In [4]: u.flush() I did the above. u.flush() was done at the end in both the shells. Final value of visit_counter is 1 (initial value is 0). I'll check on alberto's suggestion. thanks, -tml On Mar 15, 5:04 am, Alberto Valverde <[EMAIL PROTECTED]> wrote: > On Mar 15, 2007, at 1:08 AM, Andrew Grover wrote: > > > > > > > On 3/14/07, tml <[EMAIL PROTECTED]> wrote: > >> I'm using SA and have a visit counter for user profiles and some > >> other > >> pages. I have a visit_counter column in User object, and usually do > >> User.get(id).visit_counter+=1 > > >> But I realize this is a race if all the threads are serving the same > >> page. Can someone tell how to do this right? > > > If your database implements ACID features (pretty much all?) then it > > will ensure that the race is prevented. > > > You can test this by opening two tg shells -- you will see that if > > they both access the same row, the second one will freeze until the > > first one's transaction ends. > > I don't think that it would turn out this way unless some sort of > "SELECT FOR UPDATE" statement is issued by the orm. The possibility > exists of two threads reading "5", in differerent transactions, and > commiting back "6", effectively counting only one visitor. Or maybe > throwing a non-serializable update error (if using postgres). > > The table should be locked from concurrent updates when the value is > read, something like: > > table.select(table.c.id==id, for_update=True) > > I'm not sure if you can do it with a mapper, better ask in the SA list. > > Alberto ____________________________________________________________________________________ Need Mail bonding? Go to the Yahoo! Mail Q&A for great tips from Yahoo! Answers users. http://answers.yahoo.com/dir/?link=list&sid=396546091 --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "TurboGears" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/turbogears?hl=en -~----------~----~----~----~------~----~------~--~---
