Create closure to do sorting.

[aldhsu]

Was hacking on #277 and realised that splitting the directories out wasn't necessary so jumped to the sorting with a created closure.

Future architecture for sorting:

1. Take flags and "compose"(not sure how this works in Rust, so I just went with a loop), sorting functions together to create the sorting function
2. Cache sorting function to reuse for the recursive calls to Core#sort

The test pass but I would like to do some performance testing and complete the second step.

[aldhsu]

Performance testing

Using the closure:

~/Code/lsd/target/release/lsd -R --group-dirs=last  1.93s user 1.47s system 89% cpu 3.802 total

Without the closure:

lsd -R --group-dirs=last  1.91s user 1.49s system 89% cpu 3.821 total

Doesn't look like much has changed.

[aldhsu]

@Peltoche this is "ready" now. Let me know what I can do to get it merged.

[Peltoche]

Hi @aldhsu ,

A big PR with a lot of change have been merge. I think we are ready now to merge your PR if you succeed to rebase it on top of the current master

[aldhsu]

Branch is up to date now.
After the refactor, it should be simpler to add new sort criteria, add support for specifying sort direction on individual fields and add support for multisorting.

Possibly minor effect on performance, recursive sorting around 100k files:
Before

Benchmark #1: lsd --recursive -r --sizesort
  Time (mean ± σ):      2.171 s ±  0.015 s    [User: 1.108 s, System: 1.037 s]
  Range (min … max):    2.149 s …  2.187 s    10 runs

After, branch built with --release

Benchmark #1: /Users/allen.hsu/Code/lsd/target/release/lsd --recursive -r --sizesort
  Time (mean ± σ):      2.261 s ±  0.032 s    [User: 1.187 s, System: 1.042 s]
  Range (min … max):    2.218 s …  2.328 s    10 runs

[Peltoche]

This is way easier to read, thanks!

I think we can improve a little the readability with an enum for the sort order

[Peltoche]

I would be great to replace the boolean with a more explicite enum like:

enum SortOrder {
  Default,
  Reverse,
}

It would give something like:

sorters.push((SortOrder::Reverse, Box::new(with_dirs_first)));

[aldhsu]

Oh yeah that totally makes sense and the enum already exists. Addressed in bab0f0e.

[zwpaper]

@aldhsu this change looks great, can u help to fix the conflict and make this PR happen?

[codecov-commenter]

Codecov Report

Merging #281 into master will increase coverage by 5.95%.
The diff coverage is 82.70%.

@@            Coverage Diff             @@
##           master     #281      +/-   ##
==========================================
+ Coverage   66.05%   72.01%   +5.95%     
==========================================
  Files          19       19              
  Lines        1803     1865      +62     
==========================================
+ Hits         1191     1343     +152     
+ Misses        612      522      -90     

Impacted Files	Coverage Δ
src/app.rs	73.33% <ø> (+36.49%)	⬆️
src/core.rs	0.00% <0.00%> (ø)
src/display.rs	26.39% <19.04%> (-0.28%)	⬇️
src/meta/mod.rs	18.62% <62.50%> (+1.51%)	⬆️
src/icon.rs	97.85% <80.95%> (+1.34%)	⬆️
src/meta/name.rs	93.04% <89.07%> (-3.23%)	⬇️
src/meta/size.rs	89.38% <91.66%> (+22.39%)	⬆️
src/meta/filetype.rs	82.10% <92.00%> (+4.99%)	⬆️
src/sort.rs	96.11% <92.45%> (+27.12%)	⬆️
src/color.rs	79.41% <100.00%> (+6.43%)	⬆️
... and 16 more

---

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 8e1662d...05a6998. Read the comment docs.

[aldhsu]

@zwpaper I've updated the PR. I have a bit more knowledge about Rust now and I realized the closure is actually not necessary since closures are just implicit structs. I switched it to store a vec of function pointers on the Core struct instead of a closure, I think this is a bit easier to work with.

Before

Benchmark #1: lsd -R --group-dirs=last
  Time (mean ± σ):     885.4 ms ±  26.4 ms    [User: 551.8 ms, System: 318.1 ms]
  Range (min … max):   854.0 ms … 938.9 ms    10 runs

After

Benchmark #1: ~/Code/lsd/target/release/lsd -R --group-dirs=last
  Time (mean ± σ):     923.2 ms ±  23.5 ms    [User: 596.3 ms, System: 318.9 ms]
  Range (min … max):   909.3 ms … 988.2 ms    10 runs

This still definitely makes sorting a bit slower.

[zwpaper]

@meain @Peltoche could u help to review this PR, this could help developing sorting functionality a lot

[meain]

I will try to take a look at it some time this week.

[meain]

Hey, thanks a lot. LGTM.
Sorry for the delay.