Enable noUncheckedIndexedAccess

[krassowski]

References

A number of errors in the codebase such as #15103 (comment) could be prevented by enabling noUncheckedIndexedAccess.

Trying to evaluate the benefit/cost ratio in this PR.

Related to #16038

Code changes

TBD

User-facing changes

None

Backwards-incompatible changes

None

[jupyterlab-probot]

Thanks for making a pull request to jupyterlab!
To try out this branch on binder, follow this link:

[krassowski]

There is a few annoyances here:

val: string
const [key, value] = val.split('=');

both key and valueare treated as string | undefined although key is in fact string.

We cannot use parts.length > 1 for a check, but need to do parts[0] !== undefined:

export function join(...parts: string[]): string {
    if (parts[0] === undefined) {
      throw Error('join requires at least one string part')
    }
    parts[0]
}

Changing typing to:

export function join(...parts: [string, ...string[]]): string {
    parts[0]
}

works but requires calling with first ad rest like:

// cannot do this: join(...url.split('/').map(encodeURIComponent));
const [first, ...rest] = url.split('/').map(encodeURIComponent);
join(first as string, ...rest);

again because it cannot infer that first will be always string