Entire document is fetched even when no hooks are registered

[ephemer]

Short explanation: The line here https://github.com/matb33/meteor-collection-hooks/blob/master/update.js#L27 checks whether aspects.before or aspects.after exist. This check is incorrect though, because aspects.before and .after are arrays, which are always 'truey'.

What this means in practice is that every time any update occurs on any document in any collection (even without any hooks being registered), that entire document is fetched and then thrown away again by the collection-hooks package.

This came up in our Kadira performance monitoring, because some of our documents are over 1MB in size, meaning updates to that collection took up to 1s in production for just an indexed query + simple increment operation. We narrowed the issue down to this package, which we have subsequently removed (we didn't need its functionality). But I thought I'd post an issue here in case others are wondering, and to kindly suggest a fix for package users (I'd check the array lengths).

After removing the package our updates take <1 ms as they should.

[namirsab]

Thanks for posting the issue here. If you like you can write a pull request. If not ill take a look as soon as i can.

[ephemer]

I'm going to leave this one to you. Have a good weekend!

[zimme]

This should be solved in the next version.