[BUG] Open-Meteo weirdness (confirmed)

[rhester72]

Description

I consistently receive this error when updating via Open-Meteo:

Error: Received Incomplete data. Details: Warning: Less than 7 days of daily forecast data, Using Incomplete data

Steps to Reproduce

Set the following values in $HOME/.config/pi-inky-weather-epd.toml:

[api]
latitude = 40.631875
longitude = -74.437935
provider = "open_meteo"

(The above presumably uses NOAA data being in a US location.)

Expected Behavior

7 days of forecast data is retrieved.

Actual Behavior

Only 6 days of data are consistently retrieved.

Environment

Version 0.7.0 on Trixie 64-bit on a Raspberry Pi Zero 2 W

Additional Context

Thinking I could have hit a request limiter, I tried fetching the same data from their web UI (using the same source IP address) and consistently received 7 days of forecast data (as the API says should be available and defaults to).

[rhester72]

Turns out it must have been some sort of limiter, because exactly 24 hours after I started hammering Open-Meteo via Inky-Weather, I now have a next-Wednesday forecast. Closing unless it repeats.

[rhester72]

Unfortunately, today the problem is back and Thursday is N/A. No changes made whatsoever.

[mt-empty]

I will review the logs on mine.

Meanwhile if you see the same issue, could you attach the json files under ./cached_bom_data/ (bom in the name is a typo )

[mt-empty]

Ok, I think I replicated the bug. it is an off by one issue between UTC and your local time, will submit a fix in the next few days

[rhester72]

Is the fix in pull #21 complete but requiring testing or incomplete? (Not sure whether I should build and test or not yet.)

EDIT: I optimistically decided to test and learned to compile Rust. 1 out of 1 runs worked! LOL Will keep an eye for several days.

[mt-empty]

Check the alpha release https://github.com/mt-empty/pi-inky-weather-epd/releases/tag/v0.8.1-alpha.1

[rhester72]

Please note that the days printed in the future forecast are now off by one (i.e. Tue should be Wed, Wed should be Thu, et al)

Determined by comparison against prior screenshots from the site and the fact that the forecasted temperature for tomorrow is WAY higher than the high was today =)

The forecast is correct, it's the day labels that are off by -1

[rhester72]

I wonder if this is a matter of what 'tomorrow' (the day, not the date) is with respect to UTC. Will know in an hour.

EDIT: Not what I thought would happen. Sure enough, at UTC 0:00 (7:00PM local time), the weekly forecast shifted left, so now Wednesday is 'tomorrow' (as it is, for the next 5 hours)...but the forecasts shifted with it, so now Wednesday's forecast is wrong.

There's definitely a (significant) disconnect between local time and UTC with calculations.