Dylan Thacker-Smith

Yes, I understand how this makes sense in Jekyll. > I don't know if we have a good way forward since we do want to allow liquid-c compatibility here. I'm...

The embedded libmemcached has been upgraded to 1.0.18 on master now, which is the newest release, so I think this can be closed now.

> I can't change my cache_index to cache_index :widget_id, :widget_type because fetch_multi_by methods are only generated for cache_indexes with only one field. Right, there isn't cache_attribute by multiple columns support,...

Note that recursively embedded associations don't even fallback to loading from the cache, which makes the API even more confusing. That also means that if we introduced `cache_fetch_#{association_name}` methods for...

First of all, we should probably have a way to completely opt-out of `nil` caching for a cache key (e.g. the primary cache key for a model). Caching `nil` was...

> My apologies, I made a mistake in my proposal wording. The goal was to persist an `idc_do_not_cache_nil` into the cache for the record in the same way that deleted...

> By including the module, is Foo and its attributes being stored into cache whenever its called? It is a read-through cache, so it won't store Foo records in the...

> For my use case when this happen it actually make more sense to raise an exception to prevent the transaction from finishing with the wrong user. I would recommend...

I'm not sure when this would actually happen. Normally I would expect cache corruption to just result in stale data being in the cache. However, I'm not sure why there...

Thanks for the insight. I think we could probably check for that in the IdentityCache.fetch block in order to raise an exception before the data is put in the cache...