vesln / b Goto Github PK

npm install b brings in 2.0.1 but package.json says 3.0.0 on github.
Is 3.0.0 ready to try out in anger?

Getting this error:

Error: Cannot find module 'when/decorate'
at Function.Module._resolveFilename (module.js:338:15)
at Function.Module._load (module.js:280:25)
at Module.require (module.js:364:17)
at require (module.js:380:17)
at Object. (/home/david/Private/hub/common/qlobber/node_modules/b/node_modules/foreach/series.js:2:16)
at Module._compile (module.js:456:26)
at Object.Module._extensions..js (module.js:474:10)
at Module.load (module.js:356:32)
at Function.Module._load (module.js:312:12)
at Module.require (module.js:364:17)
Exited with code: 8.

I think the npm dedupe is removing the wrong one.

node_modules/b/node_modules/when/package.json says it's version 2.2.1

But

node_modules/b/node_modules/foreach/pakage.json says it needs 3.2.4 tarball

being able to write several benchmarks in a single file would make it a lot cleaner to generate a series of benchmarks with only slight differences. e.g. by using higher order functions and that sort of thing. Initially I though this would make running benchmarks in separate processes hacky and leaky but on second thought I can't see why that needs be the case. So long as there aren't any major side effects while loading a suite.

I think fixed iteration and fixed time benchmarks will both be affected by hard to predict platform level performance fluctuations. e.g. if you run the same file in a comparison against itself the second instance will be faster 9 times in 10. I guess the hardware warms up or something. A way around this would be to take multiple samples until the sample error falls below a certain level. statistics theory required though!