ttl of zero about cache-manager

[BryanDonovan]

Good point.. I could at least add that. Might take a few days though. Thanks!

[BryanDonovan]

This appears to be how the underlying lru-cache module works. This line in particular checks if a cached item is stale: https://github.com/isaacs/node-lru-cache/blob/master/lib/lru-cache.js#L422

.. it sees that the maxAge (the ttl) is 0 and returns false. I'm not sure if this is intentional or not.

[BryanDonovan]

p.s., I upgraded node-cache-manager to use the latest lru-cache in version 1.4.0. Older versions of lru-cache behaved the same way.

I also updated this line in the memory store:
ec114bc#diff-774c96972f8630e6e67be1ca15041a00R10

.. it wasn't passing the 0 ttl to lru-cache. This wasn't affecting you since you were passing the ttl in the wrap function, which was correctly getting passed to lru-cache here: https://github.com/BryanDonovan/node-cache-manager/blob/master/lib/stores/memory.js#L25

[simlu]

Just to clarify, the current behavior for 0 is now to not cache, correct? Can this be documented please?

[BryanDonovan]

@simlu - The problem is that it's not an issue with node-cache-manager, it's just how lru-cache works.

[simlu]

So passing "0" actually never expires then? I'm ok with either behaviour. Just would prefer if it was documented

[BryanDonovan]

I'm not 100% sure actually. You'd have to test how it works with whatever cache store you're using. node-cache-manager works with three different redis libraries, two mongo libraries, lru-cache, three file system caches, a hazelcast store, memcache store, and a memory store. I unfortunately can't document how they all work. node-cache-manager simply passes the 0 to the underlying store (or at least that's what it's supposed to do; if not that's a bug in node-cache-manager).

[simlu]

Ah! That makes a lot of sense. In that case should be documented that zero behaviour depends on underlying data store and should be avoided (ie prefer using 1ms)