Performance guidelines

The translations of each model is stored in a separate table. In some cases, this may cause in N-query issue. django-parler offers two ways to handle the performance of the dabase.

Caching

All translated contents is cached by default. Hence, when an object is read again, no query is performed. This works out of the box when the project uses a proper caching:

CACHES = {
    'default': {
        'BACKEND': 'django.core.cache.backends.memcached.MemcachedCache',
        'KEY_PREFIX': 'mysite.production',  # Change this
        'LOCATION': '127.0.0.1:11211',
        'TIMEOUT': 24*3600
    },
}

You have to make sure your project has the proper backend support available:

pip install python-memcached

Now, the translation table only has to be read once per day.

Query prefetching

By using prefetch_related(), all translations can be fetched in a single query:

object_list = MyModel.objects.prefetch_related('translations')
for obj in object_list:
    print obj.title  # reads translated title from the prefetched queryset

Note that the prefetch reads the information of all languages, not just the currently active language.

When you display translated objects in a form, e.g. a select list, you can prefetch the queryset too:

class MyModelAdminForm(TranslatableModelForm):
    def __init__(self, *args, **kwargs):
        super(MyModelAdminForm, self).__init__(*args, **kwargs)
        self.fields['some_field'].queryset = self.fields['some_field'].queryset.prefetch_related('translations')