Prompt for reason for skipping. #520
Conversation
| @@ -37,7 +37,8 @@ const handler = async (req, res) => { | |||
| }); | |||
|
|
|||
| // Update the backend with our Task ID | |||
| await oasstApiClient.ackTask(task.id, registeredTask.id); | |||
There was a problem hiding this comment.
Controversial proposal: What if we really delete this here and move the auto-ack to when we submit the answer?
There was a problem hiding this comment.
in that case, shouldn't the auto ack happen in the (real) backend when we submit?
I don't see the advantage of acking & submitting immediately.
TBH, I still don't fully understand why we ack in the first place.
There was a problem hiding this comment.
I agree the acking is a bit weird. Let's discuss in the next call about how needed it really is. For now let's pretend its required and just work around the API as is.
website/src/pages/api/reject_task.ts
Outdated
| import { oasstApiClient } from "src/lib/oasst_api_client"; | ||
|
|
||
| /** | ||
| * Returns a new task created from the Task Backend. We do a few things here: |
There was a problem hiding this comment.
You'll want to delete these comments
Note: skipping is only half implemented for labelling and summarisation tasks as those probably need to be changed to use the new Task component.
othrayte
force-pushed
the
skip-with-reason
branch
from
562595d to
f277e2d
Compare
|
I've moved the ACK to just before submitting interactions. Also I've fixed the summerisation and labeling tasks temporarily by making them skip without NACK-ing as they don't yet use the new Task.tsx component and I wasn't sure if someone was already working on doing that and figured that would be the better way to handle them. |
othrayte
requested review from
AbdBarho and
fozziethebeat
and removed request for
AbdBarho and
fozziethebeat
|
Apologies for any review request spam, I'm still learning how that action works on github. |
|
Looks good, we can fix any small issues later. |
This is a draft for #310 (Add ability to tag tasks as unanswerable).
It tasks onboard the feedback that we want to guide people away from skipping a bit and that we want to know why they have skipped. The leading idea for that appeared to be allowing the user to provide a custom reason, so that is the direction this went.
The main reason that this is only a draft PR is that at the moment we are not able to NACK after ACK-ing and the website auto-ACKs on fetching tasks so there is no chance to NACK. In this PR I have commented out the auto-ACK-ing to allow for testing this.