add expiration_time to cache

[vera1118]

Hi, I want to use your great package, but I need a feature to deal with expiration for cache elements. So I added expiration_time to alru_cache and wrote a simple test. Do you want to update the cache? Let me know if you want and feel free to give suggestions. Thank you!

[codecov[bot]]

Codecov Report

Merging #50 into master will decrease coverage by 2%. The diff coverage is 88%.

@@         Coverage Diff          @@
##           master   #50   +/-   ##
====================================
- Coverage     100%   98%   -2%     
====================================
  Files           1     1           
  Lines         136   150   +14     
  Branches       24    28    +4     
====================================
+ Hits          136   147   +11     
- Misses          0     2    +2     
- Partials        0     1    +1

Impacted Files	Coverage Δ
async_lru.py	98% <88%> (-2%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update e58a868...e292b73. Read the comment docs.

[hellysmile]

Hey, it mostly should be separate package cuz lru is not designed for that.

My suggestion is just each N seconds clear lru at all, if You need some kind of ttl

Also I've seen https://github.com/krkd/aiottl , which looks promising for ttl cache, but it is seems not fully ready yet

[Artimi]

Hi @hellysmile, I wanted to ask whether you would accept this PR if I edited your suggestions? Because https://github.com/aio-libs/async_lru/pull/50#issuecomment-400426038 it seems that you don't want this to be in your package (and you have all rights to say that). But I can see a use case for that, especially in async world. We are using async_lru as a cache for multiple coroutine call. We have a backend application that retrieves data from a database, does some computation and then sends it to frontend. It may happen that some request generates multiple calls of a coroutine with the same arguments and it's handy when I can just place an alru_cache decorator to save resources. However, data in the database are being updated and I need to invalidate the cache after some time. Now, easiest would be to add just expires keyword argument that would invalidate cache for me. Maybe I could have a loop that invalidates the cache for me but it's now used on several places and I think it would be better to put it right into aynsc_lru package. What do you think? Will you accept the PR if I finish it?

[Artimi]

I created a new PR https://github.com/aio-libs/async_lru/pull/131

[asvetlov]

In my mind async_lru follows functools.lru_cache design, this is the very strong point. lru_cache has no functionality for time-based expiration. Why should we have it in async version of lru_cache?

I agree with @hellysmile If you want ttl based expiration you need another library.

[Artimi]

It would just suit me very well if async_lru would have the time-based expiration. The other library you are talking about will have all the features of async_lru plus expires so I will probably resolve this by having my fork of this library. I cannot tell why lru_cache does not have this functionality but it certainly is one of the valid strategies: https://en.wikipedia.org/wiki/Cache_replacement_policies#Time_aware_least_recently_used_(TLRU). Also, alru_cache already has more arguments / method than lru_cache, e.g. cache_exceptions argument or invalidate method. If you do not want to have it the library you manage I respect it and will release my own package.

[asvetlov]

@hellysmile ?

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[asvetlov]

Implemented by c46ada7