feat: dont reorganize __cache if unbounded

[BobTheBuidler]

What do these changes do?

If a cache wrapper does not have a maxsize, there isn't any reason to reorganize the OrderedDict.

Let's wait until the codspeed stuff is merged before we look at this one.

Are there changes in behavior for the user?

None.

Related issue number

N/A

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes

[BobTheBuidler]

Since it's an internal-only attribute, can we use a builtins.dict instead of the OrderedDict in cases where maxsize is None? I believe that was the only blocker for implementing #144

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.47%. Comparing base (3341173) to head (1d96a4b).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #696      +/-   ##
==========================================
- Coverage   97.22%   96.47%   -0.76%     
==========================================
  Files          12       12              
  Lines         793      794       +1     
  Branches       41       42       +1     
==========================================
- Hits          771      766       -5     
  Misses         20       20              
- Partials        2        8       +6     

Flag	Coverage Δ
unit	95.84% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[codspeed-hq]

CodSpeed Performance Report

Merging #696 will not alter performance

_{Comparing BobTheBuidler:no-reorder-if-unbounded (9539697) with master (3341173)}

Summary

✅ 39 untouched

[Dreamsorcerer]

The benchmarks on this one look like there's a possibility this actually reduces performance slightly (I tried rerunning the benchmarks to check if it's random, but codspeed doesn't seem to update with the latest upload).

The other PRs also seem to show no discernible change in performance.

[BobTheBuidler]

You can force codspeed to rerun and update the dashboard with an empty commit, they use the commit hash as the key so if you run it twice on the same commit nothing happens iirc

The other PRs also seem to show no discernible change in performance.

I did notice this as well, though when I was looking at the flamegraph data I noticed that there's substantially more time taken by asyncio than by async-lru. I'd like to leave these open for now and think about how to best reduce the amount of asyncio stuff in the benchmark

probably by prebuilding futures and waiting until after the benchmark to deallocate them, among other potentials