On 11.07.2007., at 19:27, David Cramer wrote:
> That's an interesting solution. My main reason for not doing that is
> because I've been told the dispatcher sucks (it's slow) and we're
> going for speed.
Using CachedModel as a base class will be ideal, but ... this will
not work until model su
On 7/11/07, David Cramer <[EMAIL PROTECTED]> wrote:
>
> That's an interesting solution. My main reason for not doing that is
> because I've been told the dispatcher sucks (it's slow) and we're
> going for speed.
>
> I don't see the reasoning for adding QUERY_ to the cache_key. By
> default the cach
That's an interesting solution. My main reason for not doing that is
because I've been told the dispatcher sucks (it's slow) and we're
going for speed.
I don't see the reasoning for adding QUERY_ to the cache_key. By
default the cache_key is your db_table. So if your model is
myapp.HelloWorld it
On 11.07.2007., at 14:25, Nebojša Đorđević wrote:
> Full code is at http://dpaste.com/hold/14105/
Here is my final take on this problem: http://dpaste.com/hold/14122/
I changed caching so that all of the queries related to the one model
are stored inside single cache key. This way I can inval
This is a rough version of our cache invalidator, including a sample
decorator and a cached template tag:
http://dpaste.com/14107/
If you have multiple web servers talking to one cache (which is our
case) you can use some message passing system (like activeMQ) to
propagate the signals to other we
On 09.07.2007., at 01:11, David Cramer wrote:
> Here's about as final as we're going to get it for now: http://
> dpaste.com/hold/13884/
Great work :)
About cache invalidation ... why you can use something like this
http://dpaste.com/hold/14104/
and just add:
class Foo(models.Model):
Here's about as final as we're going to get it for now:
http://dpaste.com/hold/13884/
On Jul 5, 4:07 pm, "Jeremy Dunck" <[EMAIL PROTECTED]> wrote:
> On 7/5/07, Honza Král <[EMAIL PROTECTED]> wrote:
>
> > We haven't tested it for performance (we are building a high-volume
> > site) yet and we sti
On 7/5/07, Honza Král <[EMAIL PROTECTED]> wrote:
> We haven't tested it for performance (we are building a high-volume
> site) yet and we still haven't figured out how to deal with multiple
> web servers connecting to one cache (our working version includes
> propagating the post_save signal via s
Sorry, new link: http://dpaste.com/hold/13668/
On Jul 5, 11:41 am, "Honza Král" <[EMAIL PROTECTED]> wrote:
> On 7/5/07, David Cramer <[EMAIL PROTECTED]> wrote:
>
>
>
> > Ya cache invalidation is something you'll always have a problem with.
> > The clean() method can be used (at the end of a query
On 7/5/07, David Cramer <[EMAIL PROTECTED]> wrote:
>
> Ya cache invalidation is something you'll always have a problem with.
> The clean() method can be used (at the end of a queryset, or on the
> Manager itself) to force the invalidation.
We have been working on this issue recently and came up wi
On 7/5/07, Adrian Holovaty <[EMAIL PROTECTED]> wrote:
> Oooh, this is something I've wanted to do for ages...See
> http://code.djangoproject.com/ticket/5 , which was marked as wontfix
> by Jacob (but I'd still like to see this feature).
I'd also be interested to get input from the SoC student who
On 7/5/07, David Cramer <[EMAIL PROTECTED]> wrote:
> So I had a cool little idea -- to build a CacheManager, which let our
> team collaborate with caching controls and not duplicating data in the
> cache.
>
> So, here's a rough draft that i hammered out in 15 minutes.
>
> http://www.davidcramer.ne
Ya cache invalidation is something you'll always have a problem with.
The clean() method can be used (at the end of a queryset, or on the
Manager itself) to force the invalidation.
As for _get_sql_clause I guess I could make it just pull from filters/
extra args to build the key -- not as clean b
On 7/5/07, David Cramer <[EMAIL PROTECTED]> wrote:
> http://www.davidcramer.net/code/50/django-cachemanager.html
>
> Criticism welcomed!
I think the order of bits returned from _get_sql_clause is dependent
on how the queryset is built up, so that you'll cache equivalent
result sets repeatedly.
A
14 matches
Mail list logo
