Change wait_on_entry_calc to use while loop instead of recursion

[ericmjl]

@shaypal5 I'm submitting this PR to hopefully make cachier a bit more usable with long-running functions.

I had noticed that the recursion pattern in wait_on_entry_calc was causing the issues I was seeing in #24. I changed the implementation here, and ran a few test cases on a project I was working on.

For short-running functions, nothing breaks. The cache gets loaded correctly, and the function executes correctly. For functions that are already cached, I made sure to use overwrite_cache=True to force overwriting of the cache. For long-running functions, the function just keeps on executing nicely until it finishes up and pickles down the result. I take this back - I am still measuring to see whether this code works properly.

UPDATE 3:52 PM: I think this is working as expected. It's a long-running database query that we cache locally. I can see in pgadmin that the data are being pulled.

If you see that this is an erroneous PR, please let me know. Otherwise, please go to town with your critiques, and let me know what you'd like fixed up, I am happy to continue working on the PR till you're satisfied 😄.

[codecov]

Codecov Report

Merging #25 into master will decrease coverage by 0.63%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #25      +/-   ##
==========================================
- Coverage   95.67%   95.04%   -0.64%     
==========================================
  Files           4        4              
  Lines         324      323       -1     
  Branches       30       30              
==========================================
- Hits          310      307       -3     
- Misses          8        9       +1     
- Partials        6        7       +1     

Impacted Files	Coverage Δ
cachier/pickle_core.py	93.65% <100.00%> (-1.63%)	⬇️

---

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 4f6aff9...65a7e51. Read the comment docs.

[ericmjl]

Update on this: I can confirm that the updated cached function works as expected, even with this change made in the source.

What would you advise in terms of testing and documentation? Happy to conform to your practices here.

[shaypal5]

This is awesome, @ericmjl ! I'll take a look at this soon.

[shaypal5]

Well, for starters, the tests don't pass on Mac, so you should start with that:
https://travis-ci.org/github/shaypal5/cachier/builds/681629908

[shaypal5]

You can always see tests results on PRs here:
https://travis-ci.org/github/shaypal5/cachier/pull_requests

[ericmjl]

Oh, that's a surprise - for whatever reason, the Travis tests don't show up on GitHub:

Are they available to non-collaborators to view on the GitHub PR page?

[ericmjl]

I will take a look at the tests on macOS. That is surprising, I don't see much information available to debug:

tests/test_general.py::test_information 
cachier version: 1.4.2.post.dev7
PASSED
tests/test_general.py::test_max_workers PASSED
tests/test_general.py::test_get_executor PASSED
tests/test_general.py::test_set_max_workers PASSED
tests/test_mongo_core.py::test_information 
pymongo version: 3.10.1
PASSED
tests/test_mongo_core.py::test_mongo_index_creation PASSED
tests/test_mongo_core.py::test_mongo_core PASSED
tests/test_mongo_core.py::test_mongo_stale_after PASSED
tests/test_mongo_core.py::test_mongo_being_calculated PASSED
tests/test_mongo_core.py::test_mongo_write_failure PASSED
tests/test_mongo_core.py::test_mongo_clear_being_calculated PASSED
tests/test_mongo_core.py::test_stalled_mongo_db_cache PASSED
tests/test_mongo_core.py::test_stalled_mong_db_core PASSED
tests/test_mongo_core.py::test_callable_hash_param PASSED
tests/test_pickle_core.py::test_pickle_core Entry found.
Cached result found.
And it is fresh!
PASSED
tests/test_pickle_core.py::test_stale_after PASSED
tests/test_pickle_core.py::test_stale_after_next_time PASSED
tests/test_pickle_core.py::test_overwrite_cache PASSED
tests/test_pickle_core.py::test_ignore_cache PASSED
tests/test_pickle_core.py::test_pickle_being_calculated PASSED
tests/test_pickle_core.py::test_being_calc_next_time PASSED
tests/test_pickle_core.py::test_bad_cache_file PASSED
tests/test_pickle_core.py::test_delete_cache_file 
No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#Build-times-out-because-no-output-was-received
The build has been terminated

Given your familiarity with the codebase, do you have a hypothesis as to what might be going on?

[shaypal5]

@ericmjl I'll try and take a look soon

[shaypal5]

Ok, I see that tests/test_pickle_core.py::test_delete_cache_file is the test that gets stuck.

It's a VERY complicated test; I really can't remember it exactly, so this will require someone really digging in.

[shaypal5]

Ok, I get it. All this code sets up a very specific scenario, that makes sure that cachier correctly throws a KeyError if a function-specific cache file that it knows should exist is deleted "from under his legs".

So one thread calls a function and then deletes the cache, and second thread is called so it calls the same function and expects a cache entry and then gets the error. I don't remember/understand the exact setup now, when I look at the code, but we can make this happen.

[ofirnk]

To those who subscribed, shared a snippet for monkey-patching the fix:
#62 (comment)

[shaypal5]

Released in v1.5.2. :)