time_ns() vs time() on ESP32-S2

[fcolmant]

Hello,

I'm developing on the ESP32-S2 and encounter a strange behaviour:
The code:

import time
while True :
    s_time = time.time()
    ns_time = int(time.time_ns()/1e09)
    print('time():')
    print(s_time)
    print('time_ns():')
    print(ns_time)
    print('delta:')
    print(s_time - ns_time)
    time.sleep(1)

When I run this code after a restart, time.time_ns() is frozen. EG:
>>> %Run -c $EDITOR_CONTENT time(): 730130955 time_ns(): 730130944 delta: 11 time(): 730130956 time_ns(): 730130944 delta: 12

When I run it after a brownout, time.time_ns() is correctly implemented:
time(): 2 time_ns(): 2 delta: 0 time(): 3 time_ns(): 3 delta: 0 time(): 4 time_ns(): 4 delta: 0

Any idea?

[robert-hh]

That's strange, since both calls used the same call to gettimeofday(). Only time_ns() adds the µs value as well.. Try running it without Thonny.

[fcolmant]

Good idea, which I tried with REPL over serial. No success however.

[robert-hh]

No problem here with a ESP32S2, ESP32, ESP32C3. The above code, pasted with Ctrl-E, Ctrl-D shows a proper time series after boot. E.g.:
time():
10
time_ns():
10
delta:
0
time():
11
time_ns():
11
delta:
0
time():
12
time_ns():
12
delta:
0