Upgrade to v3 about ttlcache HOT 9 CLOSED

ok, the bolt example looks very good!

from ttlcache.

Regarding the SimpleCache question: As an outsider shopping around for caches and reading your README and godoc, I don't see what SimpleCache provides. I can see that Cache implements it. Am I also supposed to implement it if I wanted... something? A different backend? But then I'd just use a different cache.

Unless I'm missing something, I don't think it's useful. (If it's just intended as a tidy collection of the available methods, it's not needed.)

from ttlcache.

The v3 has been released.

from ttlcache.

Being forced to move away from the ReneKroon cache now. It's quite a hassle changing my code to work with this repo. I wish there was a SimpleCache.

from ttlcache.

These sound like solid improvements. Maybe leave the SimpleCache in the front example so that it's easy for people to get started?

from ttlcache.

Do you mean leave the SimpleCache interface only as an example in the README? If so, I don't think that would give much value, since all the exported methods can be seen on the godoc page. Knowing those methods, it's not hard to design your own interface that contains only the methods you need (e.g. only Set and Purge).

from ttlcache.

I do think it lowers the entry barrier for people looking 'just a cache with TTL' to seconds instead of having to study the docs. I think it's an easy way to improve exposure of the library.

from ttlcache.

Sooner or later one would have to study the docs to know how to properly use the library. However, having a few good examples that show the basic and advanced usage, like now, is useful. I'm not suggesting these examples should be removed, I just don't see how they can be better with an interface defined in them. If anything, having too much information like this increases the confusion (what if the user doesn't want to use an interface at all?) That's why the examples should be simple and to the point: they should show the usage of the most important methods, not how they can be used with other language elements. After all, if one is using go, they should already know how things like methods and interfaces work.

What I am proposing we should have is a straightforward README file with code that is somewhat similar to bolt's: https://github.com/boltdb/bolt

from ttlcache.

I also think we should have only one Get method that returns an exported *Item (its fields would remain unexported, but their accessor methods would be exported). This way we wouldn't have to create a new GetX method each time a new piece of data is needed (e.g. ttl -> GetWithTTL). How does that sound @ReneKroon?

from ttlcache.