Caching with Magic and Hyperlambda

Cache helper slots for Magic. More specifically this project provides the following slots.

All of the above slots requires a key as its value, and/or a filter argument, as the value of the identity node for the invocation of the slot. Notice, the cache can only persist string values, implying if you want to store something else besides a string, you’ll need to correctly convert the object before storing it into your cache, and correctly convert it after extracting it again later. If you need to preserve typing information, you can store Hyperlambda strings into your cache.

[cache.set]

Invoke this slot to save an item to the cache. The slot takes 3 properties, which are as follows.

Below is an example of a piece of Hyperlambda that simply saves the value of “Howdy world” to your cache, using “cache-item-key” as the key for the cache item.

cache.set:cache-item-key
   expiration:5
   value:Howdy world

To remove the above item, you can use the following Hyperlambda.

cache.set:cache-item-key

[cache.get]

Returns an item from your cache, or null if there are no items matching the specified key. Below is an example of retrieving the item we saved to the cache above.

cache.get:cache-item-key

[cache.try-get]

This slot checks your cache to look for an item matching your specified key, and if not found, it will invoke its [.lambda] argument, and save its returned value to the cache with the specified key, before returning the created item to caller.

cache.try-get:cache-key
   expiration:5
   .lambda
      return:Howdy world

If you are using the default memory based cache, then this slot is implemented in such a way that only the first invocation towards the specified key will actually execute your [.lambda] object, allowing you to have very expensive executions to create your items, that you want to store into the cache, without experiencing race conditions, or having more than one thread actually create the item and store into the cache. Notice, this is not true if you are using an out of process cache implementation, such as the one found in the “magic.data.cql” project. This should in general be your “goto slot” whenever you want to use the cache slots in this project.

[cache.clear]

This is a shorthand slot to completely clear cache, removing all items.

cache.set:cache-item-key
   expiration:5
   value:Howdy world

cache.clear
cache.get:cache-item-key

Notice, the above Hyperlambda should not return any item in its last invocation to [cache.get] since the cache was cleared before invoking it. The slot also optionally takes a filter argument as its value, which if provided, will only delete items starting out with the specified filter. Usage example can be found below.

cache.clear:foo.bar

The above will only remove cache items having a key that starts out with “foo.bar” implying that for instance the following items will be cleared.

While an item with a key of “x.foo.bar” will not be removed. This allows you to “namespace” your cache items, and clearing out for instance all cache items matching the specified namespace in one invocation.

[cache.list]

Lists all items in cache, and returns to caller.

cache.set:cache-item-key
   expiration:5
   value:Howdy world
cache.list

This slot also supports the following optional arguments.

Internals

Internally the cache implementation is not using the MemoryCache from .Net, since this class suffers from a whole range of problems in regards to its API, such as not being able to count or iterate items, etc. Hence, the actual implementation is completely custom, and is based upon the IMagicCache interface, which by default is wired up towards its MagicMemoryCache implementation. This is a conscious choice, since first of all the IMemoryCache that .Net provides out of the box is really slow, in addition to that it is missing a lot of crucial parts expected from a mature memory based cache implementation.

If you want to access the actual cache from C# or something, make sure you use it through the dependency injected IMagicCache interface, providing you with the implementation class needed to consume it from C#. This interface has a lot of nifty methods you can use to have a robust and fast memory based cache implementation in your C# code. However, it is a memory based cache. If you need better caching features going beyond what a basic memory based cache implementation can achieve, you might want to use the out of process based cache that can be found in the project called “magic.data.cql”. The latter allows for storing cache items in Cassandra or ScyllaDB.

If you consume the IMagicCache interface from your own C# code, you can create “hidden” cache items that are only possible to retrieve from C# code and not from Hyperlambda. This allows you to create a C# only type of cache, where items added as hidden cannot in any ways what so ever be retrieved from Hyperlambda. This is useful for C# only type of cache items that you don’t want to expose to Hyperlambda code. The “magic.lambda.slots” project for instance is internally storing its dynamically created slots in the cache as “hidden” items.

Project website

The source code for this repository can be found at github.com/polterguy/magic.lambda.caching, and you can provide feedback, provide bug reports, etc at the same place.

Quality gates