On Cache Invalidation: Why Is It Hard? (2018)

Re:On Cache Invalidation: Why Is It Hard? (2018) by coldtea on Monday April 05, 2021 @09:38AM (#26698761)

>Every pure function is part of a bigger machine that's eventually mutable state, or by definition that pure function has no reason to exist

Not necessarily. The universe could be just one huge pure function.

[ Parent | on HN ]

Re:On Cache Invalidation: Why Is It Hard? (2018) by kevincox on Monday April 05, 2021 @09:58AM (#26699040)

But the universe is constantly changing. So you will need to know the specific bits of the universe that your function depends on or you will never get a hit.

[ Parent | on HN ]

Re:On Cache Invalidation: Why Is It Hard? (2018) by loopz on Monday April 05, 2021 @11:50AM (#26700553)

Is the universe changing, or just your references to it?

[ Parent | on HN ]

Re:On Cache Invalidation: Why Is It Hard? (2018) by LogicUpgrade on Monday April 05, 2021 @12:07PM (#26700751)

So the references are mutable. Gotcha.

[ Parent | on HN ]

Re:On Cache Invalidation: Why Is It Hard? (2018) by loopz on Monday April 05, 2021 @04:24PM (#26704192)

Aye, functional core, imperative shell (us!).

[ Parent | on HN ]

Re:On Cache Invalidation: Why Is It Hard? (2018) by breischl on Monday April 05, 2021 @11:20AM (#26700187)

Ha, I guess it's possible. But if that's true then it seems that some of the intermediate steps in "public static void universe()" use mutable state that can result in cache invalidation difficulties.

[ Parent | on HN ]