Use 1k buffer in FileUtils.readLines to reduce heap allocation pressure

[chrisribble]

No description provided.

[chrisribble]

JMH results with various buffer sizes:

Benchmark                                      Mode  Cnt      Score     Error   Units
OshiBenchmark.readLines1k                      avgt    5     10.945 ±   1.473   us/op
OshiBenchmark.readLines1k:gc.alloc.rate        avgt    5    985.187 ± 134.199  MB/sec
OshiBenchmark.readLines1k:gc.alloc.rate.norm   avgt    5  11296.015 ±   0.002    B/op
OshiBenchmark.readLines1k:gc.count             avgt    5     69.000            counts
OshiBenchmark.readLines1k:gc.time              avgt    5     32.000                ms
OshiBenchmark.readLines2k                      avgt    5     10.561 ±   0.053   us/op
OshiBenchmark.readLines2k:gc.alloc.rate        avgt    5   1204.201 ±   6.095  MB/sec
OshiBenchmark.readLines2k:gc.alloc.rate.norm   avgt    5  13336.015 ±   0.001    B/op
OshiBenchmark.readLines2k:gc.count             avgt    5     84.000            counts
OshiBenchmark.readLines2k:gc.time              avgt    5     53.000                ms
OshiBenchmark.readLines4k                      avgt    5     10.928 ±   0.203   us/op
OshiBenchmark.readLines4k:gc.alloc.rate        avgt    5   1521.972 ±  28.238  MB/sec
OshiBenchmark.readLines4k:gc.alloc.rate.norm   avgt    5  17440.015 ±   0.001    B/op
OshiBenchmark.readLines4k:gc.count             avgt    5    106.000            counts
OshiBenchmark.readLines4k:gc.time              avgt    5     69.000                ms
OshiBenchmark.readLines512                     avgt    5     10.273 ±   0.148   us/op
OshiBenchmark.readLines512:gc.alloc.rate       avgt    5    952.852 ±  13.679  MB/sec
OshiBenchmark.readLines512:gc.alloc.rate.norm  avgt    5  10264.014 ±   0.001    B/op
OshiBenchmark.readLines512:gc.count            avgt    5     67.000            counts
OshiBenchmark.readLines512:gc.time             avgt    5     48.000                ms
OshiBenchmark.readLines8k                      avgt    5     12.006 ±   0.063   us/op
OshiBenchmark.readLines8k:gc.alloc.rate        avgt    5   2034.124 ±  10.731  MB/sec
OshiBenchmark.readLines8k:gc.alloc.rate.norm   avgt    5  25608.017 ±   0.001    B/op
OshiBenchmark.readLines8k:gc.count             avgt    5    142.000            counts
OshiBenchmark.readLines8k:gc.time              avgt    5     95.000                ms

The gains from going below 1k appear to be very small, but the delta from the default 8k to 1k is big: from 25 kB to 11 kB per invocation.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.84%. Comparing base (80ba121) to head (89f5fbd).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2612      +/-   ##
==========================================
+ Coverage   86.23%   86.84%   +0.60%     
==========================================
  Files          30       30              
  Lines        1337     1338       +1     
  Branches      189      189              
==========================================
+ Hits         1153     1162       +9     
- Misses        103      104       +1     
+ Partials       81       72       -9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dbwiddis]

LGTM with one question