I changed dur from 5 to 1 in benchmark/fs/readfile.js (solely so that I could do 60 runs in 5 minutes rather than 25 minutes, as it's getting late over here) and then did this:

$ node benchmark/compare.js --new ./node-tomhoss --old ./node-master --filter readfile.js fs 1>out.csv 
[00:04:41|% 100| 1/1 files | 60/60 runs | 4/4 configs]: Done
$ cat out.csv | Rscript benchmark/compare.R 
                                                 improvement confidence   p.value
 fs/readfile.js concurrent=1 len=1024 dur=1           1.32 %            0.1169608
 fs/readfile.js concurrent=1 len=16777216 dur=1      -0.12 %            0.6766730
 fs/readfile.js concurrent=10 len=1024 dur=1         -0.28 %            0.7146591
 fs/readfile.js concurrent=10 len=16777216 dur=1     -0.13 %            0.8874545
$

The results don't raise any concerns. However, someone should probably run a more comprehensive benchmark comparison. (For example, I didn't do writes here.)

LGTM but I'd like @trevnorris or @addaleax to take a look also

I don’t think any of these functions would actually ever be called without arguments; and when they are, they already have the overhead of failing very loudly anyways…

@tomhoss , @Trott , @jasnell , @addaleax
So, maybe, this could be safely closed without merging?

Closing without action. Can reopen later if necessary