Question

I just heard that DLR has a three level caching strategy.. But what it is .. A simple explanation with simple example will be very helpful.

Thanks

Answer 1

This is how I understand it, the idea of the caching is to reuse expressions wherever possible to reduce the dynamic vs static overhead of dynamic expression evaluation.

imagine a dynamic expression

>> a + b

Then working this out the first time an expression/syntax tree will need to be created (if one doesn’t exist). This is of the type

if a is an int and not null and b is an int and not null then result = a + b

This is essentially a rule that can evaulated and if true the expression can be used. Hence we have a level 1 cache.

Level 2 is simillar but a more complex rule, probably along the lines of:

if a is an int and not null and b is an int and not null then result = a + b
if a is string  and b is an int then do Int.Parse(a) + b
etc...

Level 3 is more complex still.

if no expression can be found then a new expression is created and added to one of the caches (though I don’t know anything about that).

As I understand it l1 is 1 rule, l2 is about 10 rules and l3 is about 100 rules.

I got all this from reading around the subject on google.
– http://dotnetslackers.com/articles/csharp/Dissecting-C-Sharp-4-0-Dynamic-Programming.aspx
– http://msdn.microsoft.com/en-us/magazine/cc163344.aspx

and some others I cannot recall now.