Add butternut to benchmarks #527
Conversation
Codecov Report
@@ Coverage Diff @@
## master #527 +/- ##
==========================================
+ Coverage 83.22% 83.26% +0.03%
==========================================
Files 34 34
Lines 2539 2539
Branches 908 908
==========================================
+ Hits 2113 2114 +1
+ Misses 255 254 -1
Partials 171 171
Continue to review full report at Codecov.
|
boopathi
added
the
Tag: Internal
README.md
Outdated
| | **closureCompiler** | 34.77kB | 72% | 11.96kB | 60% | 2.62 | 3940.01 | | ||
| | **closureCompilerJs** | 65.41kB | 48% | 15.83kB | 47% | 2.96 | 1049.21 | | ||
| | **babili** | 35.86kB | 71% | 12.47kB | 58% | 8.14 | 1418.56 | | ||
| | **uglify** | 35.74kB | 71% | 11.97kB | 60% | 2.60 | 676.57 | |
There was a problem hiding this comment.
lets add uglify-es as well?
scripts/benchmark.js
Outdated
| @@ -128,6 +130,9 @@ class Benchmark { | |||
| const out = compile(flags); | |||
| return out.compiledCode; | |||
| } | |||
| butternut(code) { | |||
| return butternut.squash(code, { check: true }).code; | |||
There was a problem hiding this comment.
Using the check: true option is a good idea to check that Butternut is producing valid code, but it shouldn't be used for benchmarking as it causes the output to be parsed again.
Also, Butternut generates a sourcemap by default. Since Babili and Uglify don't, we should probably disable sourcemaps for Butternut too to remove that from the equation:
-return butternut.squash(code, { check: true }).code;
+return butternut.squash(code, { sourceMap: false }).code;+ Remove check:true as it re-runs + Add sourceMaps false as by default others don't generate sourcemaps, but butternut does + Update README
|
What is the rule for using bold text in the benchmark tables? Highlight best results? For example, in the first table for React: Closure Compiler has the best results for output size, but not for parse time and minify time. |
https://github.com/Rich-Harris/butternut
/cc @Rich-Harris