Consider adding a config option to return NaN instead of throwing on invalid input (restoring pre-v10 behavior)

[rbangre-halliday]

In v10, BigNumber.DEBUG was removed and the library now always throws on invalid input. Previously, with DEBUG = false (the default), invalid values like undefined or non-numeric strings would return NaN — which many codebases rely on for safe fallthrough behavior.

Would you consider adding a config option (e.g. STRICT: false) to restore the old behavior? I'd be happy to submit a PR for this if you're open to it.

[MikeMcl]

Yes, I am currently considering doing exactly that: adding a STRICT config option. I would prefer it to be true by default so I don't have to release another major version though.

[rbangre-halliday]

Yes, I am currently considering doing exactly that: adding a STRICT config option. I would prefer it to be true by default so I don't have to release another major version though.

Perfect awesome, thank you!

[utkuacikgoz]

thank you - do you have any ETA for the change @MikeMcl

[MikeMcl]

I will publish by the 23rd March, I think.

[barrymichaeldoyle]

Nudge 😁 Our team is super keen to update to latest version, but we need this pre-10 behaviour because it's far too risky in our case to trust the v9 to v10 migration.