Daniel Cousens

The question distilled > To what extent do we have database transactions within singular `mutation` queries, and if not, why not?

Unfortunate as the answer is for now, I am closing this question as a duplicate of the following https://github.com/keystonejs/keystone/discussions/7642, I have opened a feature request [here (#8084)](https://github.com/keystonejs/keystone/discussions/8084).

Fixed for the `relationship` field in https://github.com/keystonejs/keystone/pull/8049, #7352 did resolve the problem too, but the other changes there aren't ready for merge.

@rachmanov is `findMany` in some way [ineffective](https://keystonejs.com/docs/apis/query#find-many) for your usage?

@tv42 you mentioned > those values need to be set even at build time You shouldn't need to configure `DATABASE_URL` for `build`, and `SESSION_SECRET` can be configured to be randomly...

@gautamsi neat metrics! I think the memory requirements could be tracked in a different issue, but I wonder if we could CI that graphed it over time.

@gautamsi could you rebase this? LGTM

@gautamsi is there a particular reason spawning a CLI process wasn't the preferred approach to the problem that I assume this is trying to resolve?

The naming, terminology and use of 'subtraction' here isn't sensible, personally. The type of the function is `subtractFromIntervals(dates: [DateRange], operand: DateRange) => [DateRange, ...]`... that is so strange to read....

@kossnocorp that's an acceptable answer, but, let's `if ('milliseconds' in duration) throw TypeError` then?