Inconsistent timeout behavior between USB and Ethernet backends

[hperrey]

spectrometer and system information

• model: HDX and QE65000
• operating system: Rocky Linux 8.10
• python version: 3.9.16
• python-seabreeze version: 2.9.2
• installed-via: conda

current problem

In a mixed-spectrometer setup, I noticed that while the Ethernet connection on the HDX would time out after O(10)s, the USB connection to the QE65000 would not. The relevant parts of the code are:

• USB:

  python-seabreeze/src/seabreeze/pyseabreeze/transport.py

  Lines 233 to 236 in 8090c59

  	def default_timeout_ms(self) -> int:
  	if not self._device:
  	raise RuntimeError("no protocol instance available")
  	return self._device.pyusb_device.default_timeout # type: ignore

• Ethernet:

  python-seabreeze/src/seabreeze/pyseabreeze/transport.py

  Lines 503 to 509 in 8090c59

  	def default_timeout_ms(self) -> int:
  	if not self._device:
  	raise RuntimeError("no protocol instance available")
  	timeout = self._device.socket.gettimeout()
  	if not timeout:
  	return 10000
  	return int(timeout * 1000)

The value for self._device.pyusb_device.default_timeout seems to be quite different from my (arbitrary) choice of 10000ms for the default socket timeout.. This is likely also related to #146

I can configure the timeout by calling spec._dev._raw_device.socket.settimeout(timeout_in_s) directly but was wondering whether there is a less awkward mechanism in seabreeze of setting that -- and if not, how one would best introduce one :)

Some more details on my user story:
While 10s seems a reasonable default timeout for internal triggers, in my particular case, I am setting up hardware triggers which might only be generated after much longer time than 10s. So before the setup generates the first trigger, the HDX would have already raised a TimeoutError.

(From my short investigation into the state of the HDX after the timeout, I actually believe that the hardware is still stuck waiting for triggers and I am not sure at this point whether one can recover from the software side without either sending a trigger or resetting the hardware. But this is not what I wanted to raise in this issue.)

steps to reproduce

Using a USB connection:

In [1]: import seabreeze
   ...: from seabreeze.spectrometers import Spectrometer, SeaBreezeError
   ...: seabreeze.use("pyseabreeze")

In [2]: spec_usb = Spectrometer.from_serial_number("QEPB0123")

In [3]: spec_usb.trigger_mode(3)  # hw triggers

In [4]: spec_usb.intensities(0,0)  # this does not time out even after minutes

Using an Ethernet connection and the HDX:

In [1]: import seabreeze
   ...: from seabreeze.spectrometers import Spectrometer, SeaBreezeError
   ...: seabreeze.use("pyseabreeze", network_adapter="192.168.254.200")

In [2]: spec = Spectrometer.from_serial_number("HDX01234")

In [3]: spec.trigger_mode(1) # hw rising edge

In [4]: spec.intensities(0,0)  # this would time out after 20s

[ap--]

So a while ago a fix was added to pyseabreeze to extend that timeout dependent on the integration that was set. It was combined into another PR which is why it wasn't very visible:

#223

you could probably pass the same through to the ethernet backend?