Suor/django-cacheops

A slick ORM cache with automatic granular event-driven invalidation.

Manjit Kumar

@manjitkumar

Hi, I was exploring caching options for and encountered django-cacheops on Github, seems pretty good with granule level of invalidation as satisfy my use cases.

Alexander Schepanovski

@Suor

Hi, glad you liked it

Manjit Kumar

@manjitkumar

I was able to configure it with my project where it's not working for automatic queryset caching whereas .cache() does cache the results. Any one can tell me what might be wrong.

CACHEOPS_REDIS = "redis://localhost:6379/1"

CACHEOPS_DEFAULTS = {
    'timeout': 60*60
}


CACHEOPS = {
    '*.*': {},
}

CACHEOPS_DEGRADE_ON_FAILURE = True

Alexander Schepanovski

@Suor

Did you supply list of operations to cache?

You've not

Manjit Kumar

@manjitkumar

CACHEOPS = {
    'api.Topic': {'ops': 'all', 'cache_on_save': True},       # cache all operations Topic model
    'api.Subscriber': {'ops': 'all', 'cache_on_save': True},  # cache all operations Subscriber model
    'api.SubscriberContract': {'ops': 'all', 'cache_on_save': True},  # cache all operations SubscriberContract model
}

This should be like this set ops to all

Alexander Schepanovski

@Suor

Yes

Manjit Kumar

@manjitkumar

oh i get it, apologies i misread the config in README.md. Thanks for your help :)

Alexander Schepanovski

@Suor

You are welcome

Chetan Dhembre

@chetandhembre

hi .. how can i invalidate nested function

Alexander Schepanovski

@Suor

There is no easy way

Chetan Dhembre

@chetandhembre

can you tell me?

Alexander Schepanovski

@Suor

And if your function is nested it probably means its value depends not only on arguments but on closure, so you can cache different things into same key

If this is not the case then you can just unnest it

Chetan Dhembre

@chetandhembre

i guess i will unnest it

i want to create my own key for function .. thats why i decided to go with nested function

Alexander Schepanovski

@Suor

Hmm, not sure I still follow

Why do you need custom key?

Maybe you can post the code?

Chetan Dhembre

@chetandhembre

def get_a(start, end):
    start_str = start.date().isoformat()
    end_str = end.date().isoformat()
    key = '%s-%s' % (start_str, end_str)

    @cached_as(extra=key, timeout=30 * 60)
    def _get_a():
        a = some_function_call()
        # do not want to cache results if a are empty
        if len(a.keys()) == 0:
            e = UncachedResult(a)
            raise e

        return a

    try:
        return _get_a()
    except UncachedResult, e:
        return e.result

Alexander Schepanovski

@Suor

First argument to @cached_as() should be queryset or model or object

This is why it's called "cached as"

And I don't see invalidation here

Chetan Dhembre

@chetandhembre

can i use cached?

Alexander Schepanovski

@Suor

Anyway, if you want to use invalidation you need to use simple - @cached(), and unnest _get_a() moving extra to arg

Chetan Dhembre

@chetandhembre

problem with that is my current nested function will get (key, start, end) as argument

start/end are datetime with second level granularity

how can i just use only single argument for caching purpose?

Alexander Schepanovski

@Suor

Just use start and end, and don't bother with key

@cached(timeout=30 * 60)
def _get_a(start, end):
    a = some_function_call()
    # Do cache empty results
    if not a:
        raise UncachedResult(a)

    return a

def get_a(start, end):
    start_str = start.date().isoformat()
    end_str = end.date().isoformat()

    try:
        return _get_a(start, end)
    except UncachedResult as e:
        return e.result

Chetan Dhembre

@chetandhembre

actually i missed one thing in my code here. .i am using start and end in my some_function_call() which are mostly get from datetime.utcnow()

so above your solution wont work

i need second level granularity for start and end

Alexander Schepanovski

@Suor

It will, just pass then

*them

What do you mean by second level granularity?

So it's

@cached(timeout=30 * 60)
def _get_a(start, end):
    a = some_function_call(start, end)
    # Do cache empty results
    if not a:
        raise UncachedResult(a)

    return a

def get_a(start, end):
    try:
        return _get_a(start, end)
    except UncachedResult as e:
        return e.result

Chetan Dhembre

@chetandhembre

end = datetime.now()
start = end - timedelta(days-12)

what will get key of cache result for _get_a()?

Alexander Schepanovski

@Suor

To get a key:

_get_a.key(start, end)

To invalidate:

_get_a.invalidate(start, end)

Chetan Dhembre

@chetandhembre

i want to cache result for 30 minutes ..
so result for datetime.datetime(2016, 10, 14, 12, 17, 0, 0) will be same for datetime.datetime(2016, 10, 14, 12, 3, 0, 0)

Alexander Schepanovski

@Suor

Then do not pass timestamp to _get_a() and some_func(), pass dates.

Chetan Dhembre

@chetandhembre

i can not

Alexander Schepanovski

@Suor

why?

Where communities thrive

People

Repo info

Activity