Is there way to "inline" function?

[Smajdalf-Smejdoplast]

In micropython on ESP32-S2 function calls are very expensive - like 100 us for call of an empty function. Is it posiible to "inline" a function somehow? I am writing a driver for RMT in Rx mode and I am trying to implement a FIFO. My idea is like

class RMTChannel():

  def add_to_FIFO(self, value):
    self.FIFO[pointer]=value
    self.head+=1
    if self.head>=len(self.FIFO)

  def read_next(self):
    value=some_magic_reading_raw_registers
    self.add_to_FIFO(value&0xFFFF)
    self.add_to_FIFO(value>>16)

Of course I can copy-paste the body of add_to_FIFO in the two locations but it quickly makes the code unreadable. OTOH using the "hacky" way I am able to read from RMT a pulse every 50 us or so but the readable way takes much much longer.

[dhalbert]

Would a collections.deque work for you? https://docs.micropython.org/en/latest/library/collections.html#collections.deque

[peterhinch]

See also micropython.RingIO - an efficient circular buffer.

[jonnor]

This is explicitly designed for cases like interrupts, so should be good for this case.

[GitHubsSilverBullet]

In micropython on ESP32-S2 function calls are very expensive - like 100 us for call of an empty function.

That'd be embarrassingly slow. I faced a similar situation quite a while ago and wrote a test to verify the various call penalties.
Normal Python, with native decorator, with viper decorator, and assembler.
For each type with none to four parameters.
Each version is called 10000 times and the timing then divided by 10000.
Interestingly a viper call is a tad faster than an assembler call.
Here are the results for a rp2350:

Normal  :  9.901 µs/call
Normal  : 10.155 µs/call
Normal  : 10.256 µs/call
Normal  : 10.334 µs/call
Normal  : 10.798 µs/call
native  :  4.251 µs/call
native  :  4.437 µs/call
native  :  4.616 µs/call
native  :  4.709 µs/call
native  :  5.076 µs/call
viper   :  2.916 µs/call
viper   :  3.029 µs/call
viper   :  3.142 µs/call
viper   :  3.255 µs/call
viper   :  3.388 µs/call
asm     :  2.883 µs/call
asm     :  3.109 µs/call
asm     :  3.343 µs/call
asm     :  3.569 µs/call
asm     :  3.782 µs/call