This is pretty hackish, and might not be fast enough, but if the data is simple(ie just 1 number to represent something) perhaps just store it in a text file.
On Feb 15, 12:23 am, Amit Ramon <[EMAIL PROTECTED]> wrote: > After reading the replies, and thinking about the problem a little bit more, > I think I can phrase it better (the replies really helped me to better > understand it). > > I need to draw one of, say, N elements in every request, in such a way that > over many requests the drawing distribution will be uniform. It doesn't have > to be random, though. This can be achieved either with a deterministic > mechanism, or via a probabilistic one. So the state I was talking about could > be simply an integer value. > > In the previous posts we discussed mostly deterministic mechanisms. I now > realize that the major requirement from such a mechanism is that the actions > of retrieving a value and incrementing it should be done in one atomic action > (atomic across concurrent requests). It seems that due the way Django and > FastCGI work, it is not easy to find a deterministic mechanism that will work > and also perform well. (I think that I can use a dummy model and create and > save a new instance whenever I need a new value, and use the instance id as > the value, but it seems too costly). > > What I now think is that I shall look for a probabilistic mechanism. Perhaps, > assuming requests come in a more or less uniform distribution over time, I > can use the time of the request to generate my "random numbers". For example, > take the modulo N (number of elements) of the absolute request time (say > number of seconds elapsed since a fixed point in time). I may need to think > more of the math here. > > I'd happily read any comments. > > Thank for all the replies, > > Amit > > * Marty Alchin <[EMAIL PROTECTED]> [2008-02-14 11:48 -0500]: > > > > > On Thu, Feb 14, 2008 at 11:31 AM, Alex Koshelev <[EMAIL PROTECTED]> wrote: > > > > Look at idea behind > > > this:http://code.djangoproject.com/wiki/CookBookThreadlocalsAndUser > > > That will likely fail to work across multiple requests, due to > > threading and/or multi-process issues. > > > Better yet, check out this little gem: > >http://www.djangoproject.com/documentation/cache/#the-low-level-cache... > > > -Gul --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---