Diffs over 2048 characters are truncated even when there is no performance problem

[joshgoebel]

Reference: #4638

Can this be a feature flag please so it can be disabled? I was having NO speed issues before but now my test output is entirely broken (it doesn't show the actual diff at all) because it's a bit over 2000 characters. ☹️

This arbitrary 2048 limit does not seem a good heuristic.

Originally posted by @joshgoebel in #4638 (comment)

[KsenyLa]

I have similar question - can it be disabled? Feature flag would be very helpful

[jnordberg]

Also running into this, looks like its hardcoded to 2048 chars...

[joshgoebel]

Correct. At the very least that should be configurable.

[jnordberg]

I think something like this would help in many cases (it certainly would in mine) and will be reasonably fast even for huge inputs.

// truncate beginning of actual and expected strings until first difference if they exceed maxLen
const maxLen = 2048 // TODO: make this configurable :)
const len = Math.max(actual.length, expected.length)
let p = 0
while (len - p > maxLen && actual[p] === expected[p]) {
    p++
}
actual = actual.slice(p, maxLen + p)
expected = expected.slice(p, maxLen + p)

[norla]

I have run into this several times so I took the liberty to create a pull request (see above).
@jnordberg I tried including your approach, but skipping to the first diffing character caused some issues with the ouput (not showing enough context, extra diffs at the end etc) so I decided not to use it.