recursive clock uopdate - ESP32

[guyd]

Hi,

My device, based on ESP32 with Micropython, needs a RTC to run scheduled tasks.
Using

Code: Select all

import ntptime
ntptime.settime()

RTC is updating, but I wish to update it every several time. Since it is not Linux, or Python - I wish to do so with minimum resources.
what is best for that ?

Guy

[pythoncoder]

I can see no particular resource problem with periodically running

Code: Select all

ntptime.settime()

There is probably no alternative as NTP is likely to be your only source of an accurate reference.

However you do have to be careful about updating a clock which is being used to schedule events. This is because your clock no longer increments in a monotonic fashion: it can jump forwards or backwards by an arbitrary amount. Without care this can cause events to be scheduled twice or to be omitted.

[guyd]

I can see no particular resource problem with periodically running

Code: Select all

ntptime.settime()

There is probably no alternative as NTP is likely to be your only source of an accurate reference.

However you do have to be careful about updating a clock which is being used to schedule events. This is because your clock no longer increments in a monotonic fashion: it can jump forwards or backwards by an arbitrary amount. Without care this can cause events to be scheduled twice or to be omitted.

Thank you for your answer.
since I need it as an automated process - I was wondering if using thread and sleep for 6 hours between updates, is light enough - since on top of the clock update prcoess, there is the scheduled tasks process.

Can you please explain why jumps in time can occur ? what would be the reason for that ? fail update process ?

[carsten]

However you do have to be careful about updating a clock which is being used to schedule events. This is because your clock no longer increments in a monotonic fashion: it can jump forwards or backwards by an arbitrary amount. Without care this can cause events to be scheduled twice or to be omitted.

Can this cause problems with uasyncio?

[pythoncoder]

It won't cause problems with official uasyncio as it does not use the RTC as its timebase.

But if your own code checks for a time of (say) 3.10.00 to trigger an event, you could have a problem. Say the RTC time was 3.09.55 and an update caused it to be corrected to 3.10.05 the time of 3.10.00 would never occur. A correction in the reverse direction would cause it to occur twice.

[guyd]

It won't cause problems with official uasyncio as it does not use the RTC as its timebase.

But if your own code checks for a time of (say) 3.10.00 to trigger an event, you could have a problem. Say the RTC time was 3.09.55 and an update caused it to be corrected to 3.10.05 the time of 3.10.00 would never occur. A correction in the reverse direction would cause it to occur twice.

Oh, now I get it.
TNX

[carsten]

It won't cause problems with official uasyncio as it does not use the RTC as its timebase.

But if your own code checks for a time of (say) 3.10.00 to trigger an event, you could have a problem. Say the RTC time was 3.09.55 and an update caused it to be corrected to 3.10.05 the time of 3.10.00 would never occur. A correction in the reverse direction would cause it to occur twice.

Thank you for your answer!