Compare the size of libraries in the benchmark

[stasm]

How about displaying the size of each test case, including the code pulled in from the tested library, next to the performance results? I think it might be interesting to see how performance relates to the amount of code.

Disclaimer: As the author of Goodluck, I'm obviously biased. Goodluck has been designed with size-constrained use-cases in mind, and it compares favorably to other libraries in the benchmark. I won't mind it if you think this isn't helpful.

Size computation could be done with help of a bundling tool like esbuild, for example. Here's a Bash one-liner to give a sense of the spectrum of current results.

# Precondition: npm install esbuild.
find src/cases -name "*.js" | while read casePath; do
    echo "${casePath#src/cases/}";
    npx esbuild $casePath --bundle --minify | wc -c;
done

[Output as of d043805, in bytes]

becsy/entity_cycle.js
   55517
becsy/simple_iter.js
   56098
becsy/frag_iter.js
   55558
becsy/packed_1.js
   55448
becsy/add_remove.js
   55452
becsy/packed_5.js
   56034
picoes/entity_cycle.js
    5319
picoes/simple_iter.js
    5676
picoes/frag_iter.js
    5422
picoes/packed_1.js
    5376
picoes/add_remove.js
    5243
picoes/packed_5.js
    5510
wolf-ecs/entity_cycle.js
    5916
wolf-ecs/simple_iter.js
    6553
wolf-ecs/frag_iter.js
    6012
wolf-ecs/packed_1.js
    5931
wolf-ecs/add_remove.js
    5901
wolf-ecs/packed_5.js
    6567
piecs/entity_cycle.js
    7493
piecs/simple_iter.js
    8083
piecs/frag_iter.js
    7587
piecs/packed_1.js
    7483
piecs/add_remove.js
    7480
piecs/packed_5.js
    7995
perform-ecs/entity_cycle.js
    6556
perform-ecs/simple_iter.js
    7259
perform-ecs/frag_iter.js
    6603
perform-ecs/packed_1.js
    6648
perform-ecs/add_remove.js
    6581
perform-ecs/packed_5.js
    7268
ecsy/entity_cycle.js
   22070
ecsy/simple_iter.js
   23034
ecsy/frag_iter.js
   22181
ecsy/packed_1.js
   22239
ecsy/add_remove.js
   22027
ecsy/packed_5.js
   22939
goodluck/entity_cycle.js
    1457
goodluck/simple_iter.js
    2140
goodluck/frag_iter.js
    1636
goodluck/packed_1.js
    1715
goodluck/add_remove.js
    1303
goodluck/packed_5.js
    2031
bitecs/entity_cycle.js
    9166
bitecs/simple_iter.js
    9704
bitecs/frag_iter.js
    9143
bitecs/packed_1.js
    9060
bitecs/add_remove.js
    9469
bitecs/packed_5.js
    9571
javelin-ecs/entity_cycle.js
   15524
javelin-ecs/simple_iter.js
   16127
javelin-ecs/frag_iter.js
   15600
javelin-ecs/packed_1.js
   15548
javelin-ecs/add_remove.js
   15483
javelin-ecs/packed_5.js
   15898
uecs/entity_cycle.js
    3401
uecs/simple_iter.js
    3743
uecs/frag_iter.js
    3527
uecs/packed_1.js
    3470
uecs/add_remove.js
    3363
uecs/packed_5.js
    3614
harmony-ecs/entity_cycle.js
   10585
harmony-ecs/simple_iter.js
   10919
harmony-ecs/frag_iter.js
   10622
harmony-ecs/packed_1.js
   10550
harmony-ecs/add_remove.js
   10582
harmony-ecs/packed_5.js
   10945
geotic/entity_cycle.js
   10980
geotic/simple_iter.js
   11571
geotic/frag_iter.js
   11115
geotic/packed_1.js
   11111
geotic/add_remove.js
   10940
geotic/packed_5.js
   11355
tiny-ecs/entity_cycle.js
   14711
tiny-ecs/simple_iter.js
   15291
tiny-ecs/frag_iter.js
   14831
tiny-ecs/packed_1.js
   14778
tiny-ecs/add_remove.js
   14663
tiny-ecs/packed_5.js
   14978

[EnderShadow8]

Might also be a good idea to exclude the library code to compare verbosity, although that might encourage slight golfing of cases.