Discord API Issue: Role creation call might never resolve/reject about yet-another-better-office-hour-bot

[tomli380576]

Another temporary fix is to delete the server that BOB is currently in and make a new one.
For development, disable any role creation call if the role is already created.

[tomli380576]

The Promise.race() fix is at catch-role-creation-fail branch.

If we need to catch timeout everywhere, here's the decorator solution:

function timeoutable(timeoutAfter: number, errMsg: string) {
    return (_t: unknown,
        _p: string,
        descriptor: PropertyDescriptor) => {
        const promiseToRace = descriptor.value;
        let timeoutHandle: NodeJS.Timeout;

        descriptor.value = function (this: any) {
            return Promise.race([
                // eslint-disable-next-line prefer-rest-params
                promiseToRace?.call(this, arguments),
                new Promise<never>(
                    (_, reject) => {
                        timeoutHandle = setTimeout(() =>
                            reject(new Error(errMsg)),
                            timeoutAfter);
                    }
                ).finally(() => {
                    clearTimeout(timeoutHandle);
                    console.log('timeout cleared');
                })
            ]);
        };
    };
}

Usage:

class SomeClass {

    // force reject after 5 seconds
    @timeoutable(5000, "Method timed out.")
    async methodThatCanTakeForever() {
        // ...
    }
}

[KaoushikMurugan]

Might be related to this issue, maybe not. But I think the "Ensure roles exist" function itself can be avoided being called unnecessarily every time the bot needs to check if a user has a role. Instead, we should do something similar to how we process the queue messages and store the role id (or object) in an AttendingServer class member and then use that whenever doing a role check. We should only worry about if the roles exist when starting the server and when a role is deleted

edit: also store the role ids in firebase

[tomli380576]

The issue doesn't seem to show up any more. I think we can safely ignore it for now