https://www.growthbook.io/ logo
#give-feedback
Title
# give-feedback
b

busy-horse-73824

09/08/2023, 1:25 PM
It seems like it's no longer possible to schedule experiments/experiment rules? Why has that gone? The documentation says it's possible on all rule types, including experiments
👀 1
f

future-teacher-7046

09/08/2023, 1:31 PM
I think this got missed in the recent refactor of experiment rules. We can add it back in. Long term, we want to move scheduling to the experiment itself instead of the feature rule. That way, there's a single source-of-truth for the status of the experiment and it avoids a bunch of weird edge cases (e.g. experiment is "running", but the feature rule is not active yet).
👌 1
b

busy-horse-73824

09/08/2023, 1:32 PM
Yeah definitely prefer it being in the experiment 🙂
Just ideally want to be able to say e.g. "start at midnight" rather than "start now"
also one minor thing
We almost always run experiment targeting like this
Basically because we don't just want the experiment to start at a particular time, but we only want it to affect users whose account was created from that same time
Ideally it'd be good if there was a special targeting rule so we can say like
"mojo_user_created_at": { "$gte": [experiment start] }
And then have the $gte evaluated as a proper timestamp comparison (similar to the special case version comparison)
f

future-teacher-7046

09/08/2023, 1:40 PM
I'll think about ways we might be able to support a placeholder variable for experiment start date. As for "proper date comparison", we recommend using ISO dates (YYYY-MM-DDThhmmss) in UTC everywhere. Supported in almost all languages and normal string comparisons work correctly
👌 1
b

busy-horse-73824

09/08/2023, 4:18 PM
Well yeah but it is hard to be certain that every use of the input will be normalized to UTC first
And technically e.g.
+00:00
and
Z
are the same thing but not likely to be an issue
9 Views