BEMHTML: support nested mixes

[miripiruni]

Fix #241

Instead of #280

[miripiruni]

$ node runner.js --rev1 5aad540b0d97e379372558e688ab9021d44eea9b --rev2 a2d25efb5a93bbf28c8b386dbb9f95c8dc3bc2ca --bemjson 10000 --dataPath ~/bemjson --templatePath ./template-8x.js
Test started…
{ _: [],
  h: false,
  help: false,
  rev1: '5aad540b0d97e379372558e688ab9021d44eea9b',
  rev2: 'a2d25efb5a93bbf28c8b386dbb9f95c8dc3bc2ca',
  bemjson: 10000,
  dataPath: '/Users/miripiruni/bemjson',
  templatePath: './template-8x.js',
  '$0': 'runner.js',
  repeat: undefined,
  verbose: undefined }


Wed May 10 2017 17:19:30 GMT+0300 (MSK)

Total test time:  238615.161181
./lib/compare.py ./dat-5aad5-a2d25 ./dat-5aad5-a2d25/5aad540b0d97e379372558e688ab9021d44eea9b-10000-1494426090791.dat ./dat-5aad5-a2d25/a2d25efb5a93bbf28c8b386dbb9f95c8dc3bc2ca-10000-1494426208718.dat

Percentile:  0.5
{ rev1: 7.719904,
  rev2: 7.520985,
  'diff abs': 0.19891900000000007,
  'diff percent': 2.576703026358873 }

Percentile:  0.9
{ rev1: 10.773798,
  rev2: 10.624067,
  'diff abs': 0.14973099999999917,
  'diff percent': 1.3897698843063466 }

Percentile:  0.95
{ rev1: 12.338903,
  rev2: 12.196022,
  'diff abs': 0.14288100000000092,
  'diff percent': 1.1579716608518642 }

[miripiruni]

@veged @zxqfox what do you think:

~1% tradeoff (95 percentile) on benchmark — is it OK?

[coveralls]

Coverage increased (+0.001%) to 99.687% when pulling 1692272 on issue-241__nested-mixes-2 into a2d25ef on master.

[pvdz]

@miripiruni I think any perf trade should be worth the feature being implemented. It's hard to tell whether that is the case here. Furthermore this PR introduces 11k loc and there's no way in hell anybody is going to be able to review this properly, so good luck with that.

I would give more insights but aside from this PR not really being reviewable, I simply don't know enough about this project to begin with even it it was. Perhaps @zxqfox does, though.

[pvdz]

PS. It's impressive that the 11k loc and one test actually improved code coverage by 0.001%. It's in fact so impressive that would suggest somebody take a closer look at that because it's pretty unlikely to be correct.

[miripiruni]

@qfox my apologies, I think it’s second time when I confused nickname. That’s because @zxqfox have ‘qfox’ username as corporate login…

You can ban me :)

P..S. 11K loc in bench dir. But coverage watch only lib dir.

[pvdz]

You can ban me :)

shrug that's okay :)

P..S. 11K loc in bench dir. But coverage watch only lib dir.

Ah that makes sense.

[coveralls]

Coverage increased (+0.001%) to 99.687% when pulling 5aad540 on issue-241__nested-mixes-2 into 908c67d on master.

[pvdz]

@zxqfox ^

[miripiruni]

bem-xjst@8.7.1