Hooks not executing on related table queries

[DenisBessa]

Hi, first I would like to acknowledge how incredible this library is. Congratulations on the excellent work.

One of the main features, for me, are the beforeHooks and afterHooks.

I noticed that these hooks only work when the table being queried is the main one in a query.

For example, if I have hooks on the customers table, they are executed perfectly in a query like this:

const data = await db.customers.selectAll()

However, they are apparently ignored in a query like this:

const data = await db.companies
    .select('*', {
        customers: q => q.customers,
    })
    .findByOptional({ id: '1' })

Is there any way to execute the hooks in a query like this?

If not, is there any perspective of implementing this functionality?

Thank you!

[romeerez]

Thank you for the inspiring words, I really appreciate knowing this library is useful and joyful for others!

I have two specific ideas that would require triggering hooks of relations, the first doesn't seem to be complex and the second may be too complex, the ideas exist from day 1 of this library, but couldn't make it because of other issues and time limitations.

This will be done for create/update/delete callbacks one day for sure, absolutely.

But, I never thought of a use case for querying the data, could you share your use case?
Why would you want to trigger a function every time when the data is selected from the table?

From docs:

init(orm: typeof db) {
    // data is of type `unknown` - it can be anything
    this.afterQuery((data, q) => console.log('after any query'));

The data here is unknown and can be anything because maybe it was triggered by selecting count, or even a custom SQL, so in current form it's not much useful, perhaps it is worth adding an additional functionality for your use case.

[DenisBessa]

In my use case, the table contains cached data from an external API. I want to update the old rows with the newest data before running the query.

I can partially do this using the computeAtRuntime function, but I think it is not a best practice.

[romeerez]

the table contains cached data from an external API

So the table acts as a cache, there are two universal strategies for cache:

• pull model: whenever you make a query, you fetch data from the third-party, and the response may be cached for some time;
• push model: data is loaded not in the time of querying, but ideally whenever data is changed, may be done via webhook, or periodically scraped via a cron job.

Both makes sense, supporting querying third-party at a query time may be a good addition.

I can partially do this using the computeAtRuntime function

Currently, it doesn't accept async functions, but if it was supported, would it be what you're looking for?
Why not a best practice?

As I understand "best practices", they suggest that loading data from the db is one concern, loading it from third-party is a different one, they shouldn't be coupled. Since the data is originated from a third-party API, it should be handled separately and differently.