Slow Start to Evaluation

[0x4007]

We should fix this up and do an instant response when the issue is closed as complete. Its always a bit nerve wracking that its broken again because it takes like 20 seconds to post.

Easy answer is to just do a CURL post with the comment when starting the operation.

+ Evaluating results. Please wait...

Originally posted by @ubiquity-os[bot] in ubiquity-os-marketplace/daemon-pricing#32 (comment)

[gentlementlegen]

Wouldn't this be fixed buy the way you found to quickly start Actions? Because the start message is at the top of the action before any clone or install actually happens so not sure if it can be made faster.

[0x4007]

Wow I see that now. I am not optimistic about going much faster compared to what you already did. I don't understand the length of time for the delay if the comment says it posted within a second according to the GitHub actions logs.

Perhaps with caching of some sort which is inherited with my implementation from my research but I'm not sure of the nuances because I'm borrowing that code

[gentlementlegen]

Sometimes the runners are slow to pick up the task and start, if you add the delay for the comment to be posted + the UI to poll the changes I guess it can add up to ~5 seconds. I believe without using a Worker or any sort of backend that would instantly react to events, we are dependent on this delay.

[0x4007]

The only thing I can think of is caching. There is a slim chance that this is the solution but I am not very confident.

[gentlementlegen]

Even with caching the runner can take time to be picked up for start. From my experiments with ncc the running is instant but the runner pick up still varies from 1 second to 5 minutes, which I don't think can be controlled. Let's see once we get the runs up.

[0x4007]

the runner pick up still varies from 1 second to 5 minutes, which I don't think can be controlled.

Yes I agree. The only thing we could hope for is to request a run when GitHub's infrastructure isn't so busy!