Implement --deterministic, for chaotic but repeatable color selection

[saethlin]

From the runs I've tried out with this so far it seems like the color selection is a bit biased towards darker shades, which is unfortunate. Not entirely sure what to do about that.

[codecov]

Codecov Report

Merging #190 into master will increase coverage by 0.05%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #190      +/-   ##
==========================================
+ Coverage   90.31%   90.36%   +0.05%     
==========================================
  Files          17       17              
  Lines        2219     2231      +12     
==========================================
+ Hits         2004     2016      +12     
  Misses        215      215              

Impacted Files	Coverage Δ
src/flamegraph/color/mod.rs	83.55% <100.00%> (+1.03%)	⬆️
src/flamegraph/mod.rs	96.45% <100.00%> (+0.03%)	⬆️

---

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 86302c8...e65e6e2. Read the comment docs.

[jonhoo]

Thanks, this looks pretty good! Left some comments inline. I'm not sure why the colors are generally dark, though the second hash trick I mentioned might help. I'd also like to see a new test case for this.

[saethlin]

I'd love to add some tests, but... how? The existing test suite doesn't run on its own, it seems like maybe it's missing some files?

---- collapse::dtrace::tests::test_collapse_multi_dtrace_simple stdout ----
Error: Os { code: 2, kind: NotFound, message: "No such file or directory" }
thread 'collapse::dtrace::tests::test_collapse_multi_dtrace_simple' panicked at 'assertion failed: `(left == right)`
  left: `1`,
 right: `0`: the test returned a termination value with a non-zero status code (1) which indicates a failure', /home/ben/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/test/src/lib.rs:192:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

---- collapse::dtrace::tests::test_collapse_multi_dtrace stdout ----
Error: Os { code: 2, kind: NotFound, message: "No such file or directory" }
thread 'collapse::dtrace::tests::test_collapse_multi_dtrace' panicked at 'assertion failed: `(left == right)`
  left: `1`,
 right: `0`: the test returned a termination value with a non-zero status code (1) which indicates a failure', /home/ben/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/test/src/lib.rs:192:5

---- collapse::perf::tests::test_collapse_multi_perf stdout ----
Error: Os { code: 2, kind: NotFound, message: "No such file or directory" }
thread 'collapse::perf::tests::test_collapse_multi_perf' panicked at 'assertion failed: `(left == right)`
  left: `1`,
 right: `0`: the test returned a termination value with a non-zero status code (1) which indicates a failure', /home/ben/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/test/src/lib.rs:192:5

---- collapse::perf::tests::test_collapse_multi_perf_simple stdout ----
Error: Os { code: 2, kind: NotFound, message: "No such file or directory" }
thread 'collapse::perf::tests::test_collapse_multi_perf_simple' panicked at 'assertion failed: `(left == right)`
  left: `1`,
 right: `0`: the test returned a termination value with a non-zero status code (1) which indicates a failure', /home/ben/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/test/src/lib.rs:192:5

[jonhoo]

Ah, yes, it also runs a bunch of tests from the upstream Perl flamegraph repository, which has to be checked out with

$ git submodule update --init

[saethlin]

@jonhoo The test I've added passes on my machine, I would really appreciate some assistance here. I don't know why it's not working in the CI or whether this is even the test we want.

[jonhoo]

Hmm, looking at the test output, it seems like the test run generates a different color than the committed test file. Even more interestingly, different runs of the test produce different values, which suggest that the generated color is not actually deterministic. Are you sure the test is actually being run on your computer? The failure seems to be consistent across all the different OSes.

Separately, you have to use std::u64::MAX, not u64::MAX, otherwise the code will not compile on Rust 1.40.

[jonhoo]

Ah, good catch!

[jonhoo]

Excellent, thank you!

[jonhoo]

Released in 0.10.1.