Tests should not depend on hashmap iteration order

[workingjubilee]

Hi! Your test suite depends on iterating over the stdlib's HashMap<K, V, S> but we generate random state for every map unless it overrides the S parameter, meaning each time they are built and run, you get different hashing. This is intentional to prevent common exploits. But we also would like hashing details to not slow down common uses like iteration, so when you iterate over a hashmap, you get an "arbitrary order", which means that iteration of a hashmap with S = RandomState visits keys in a different order each time.

This results in this sort of thing happening to your tests:

[INFO] [stderr]     Finished `test` profile [unoptimized + debuginfo] target(s) in 0.01s
[INFO] [stderr]      Running unittests src/lib.rs (/opt/rustwide/target/debug/deps/rwini-93436e769e484a05)
[INFO] [stdout] 
[INFO] [stdout] running 2 tests
[INFO] [stdout] test parser::test::ini_parser::test_parse ... ok
[INFO] [stdout] test data::test::test_ini_to_string ... FAILED
[INFO] [stdout] 
[INFO] [stdout] failures:
[INFO] [stdout] 
[INFO] [stdout] ---- data::test::test_ini_to_string stdout ----
[INFO] [stdout] thread 'data::test::test_ini_to_string' panicked at src/data/mod.rs:13:5:
[INFO] [stdout] assertion `left == right` failed
[INFO] [stdout]   left: "[core]\nkey:value\nname:1"
[INFO] [stdout]  right: "[core]\nname:1\nkey:value"

You may wish to consider some manner of sorting before comparing in your test.

[zerodegress]

Roger that. I'll try to improve it ASAP.