Massive performance degradation since v1.7.5

using node v12.18.3

Version 1.7.5 introduced a regression which causes a massive performance degradation.

Test code:

const fs = require('fs');

const git = require('isomorphic-git');

const dir = '.';
const filepath = 'README.md';
const ref = 'master';

const run = async (label) => {
  console.time(label); 
  
  console.time('resolveRef'); 
  const commitSha = await git.resolveRef({ fs, dir, ref });
  console.timeEnd('resolveRef'); 

  console.time('status'); 
  const status = await git.status({ fs, dir, filepath });
  console.timeEnd('status'); 

  console.time('readObject #1'); 
  const { oid } = await git.readObject({ fs, dir, oid: commitSha, filepath });
  console.timeEnd('readObject #1'); 

  console.time('readObject #2'); 
  const { object: content } = await git.readObject({ fs, dir, oid });
  console.timeEnd('readObject #2'); 

  console.timeEnd(label); 
}

const main = async () => {
  await run('total run #1');
  await run('total run #2');
}

main().catch(err => console.log(err));

I’ve tested with this repository: https://github.com/davidnuescheler/pages

Results using 1.7.4:

resolveRef: 2.276ms
status: 364.792ms
readObject #1: 2.695ms
readObject #2: 1.940ms
total run #1: 376.457ms

resolveRef: 0.672ms
status: 9.638ms
readObject #1: 2.091ms
readObject #2: 1.598ms
total run #2: 14.323ms

Results using 1.7.5:

resolveRef: 2.181ms
status: 359.206ms
readObject #1: 224.917ms
readObject #2: 228.641ms
total run #1: 819.938ms

resolveRef: 2.451ms
status: 247.968ms
readObject #1: 218.427ms
readObject #2: 229.541ms
total run #2: 698.930ms

=> The second run takes ~40-50x longer using 1.7.5.

related issue: #1237

[stefan-guggisberg]

I went ahead and tested with before mentioned repo:

v1.7.8:

resolveRef: 2.230ms
status: 400.173ms
readObject #1: 294.029ms
readObject #2: 278.865ms
total run #1: 981.091ms

resolveRef: 1.959ms
status: 231.133ms
readObject #1: 254.164ms
readObject #2: 254.196ms
total run #2: 742.023ms

resolveRef: 2.117ms
status: 230.299ms
readObject #1: 229.716ms
readObject #2: 275.966ms
total run #3: 738.623ms

resolveRef: 2.010ms
status: 231.274ms
readObject #1: 228.124ms
readObject #2: 239.396ms
total run #4: 701.301ms

resolveRef: 1.771ms
status: 241.533ms
readObject #1: 223.004ms
readObject #2: 236.289ms
total run #5: 703.025ms

exposed cache param:

resolveRef: 2.162ms
status: 352.665ms
readObject #1: 2.634ms
readObject #2: 1.923ms
total run #1: 364.962ms

resolveRef: 0.599ms
status: 2.456ms
readObject #1: 2.215ms
readObject #2: 1.469ms
total run #2: 7.018ms

resolveRef: 0.456ms
status: 1.888ms
readObject #1: 1.503ms
readObject #2: 0.670ms
total run #3: 4.739ms

resolveRef: 0.443ms
status: 2.021ms
readObject #1: 1.281ms
readObject #2: 0.485ms
total run #4: 4.443ms

resolveRef: 0.454ms
status: 1.675ms
readObject #1: 1.213ms
readObject #2: 4.168ms
total run #5: 7.717ms

It’s roughly a 100x improvement.

[stefan-guggisberg]

Thank you very much, @wmhilton. That solved our performance issue.