Ben Manes

I think that I supplied `currentTime` just because it was a known value, but not being wall clock meant that I didn’t know if it could be useful either. I...

Unfortunately any evolution of the apis here has some drawback that makes it potentially confusing. ### Cache builder If we extended the cache builder to accept lambda expressions then we'd...

Experimenting with the api a bit, and the `triFunction` to include a `currentDuration` parameter doesn't make sense. The reasons why I thought it might didn't pan out, which is a...

I think it would be fine to reorder those lines or await on the `whenComplete` future as well, as a refinement to the behavior. It makes sense that a caller...

Thanks! I'll take a look. There are a few things that I want to review while we're looking at this functionality. 1. Can we write tests to assert this behavior...

Thanks! I think that I understand what may be happening. When the entry is being loaded and the refresh duration passes since mapping's insertion, then attempts to reload should no-op....

In that case you can hack around it for the moment by doing a no-op write after the load. It's not pretty, but it will clear the entry from mapping...

sorry, I just haven't had the time and mental focus to work on this project. I'll keep chipping away and I might cut a quick release with just this fix...

I think the original fix is correct, but it also enters this unexpectedly because the write time was not adjusted far into the future. When the future is first loaded...

Thanks for taking the time to benchmark and provide your findings. Here are a few observations, 1. You should probably use jmh for benchmarks to ensure you do not bias...