Unable to load C++ Class with multiclass Python HL

# 2 years ago
alemeillour	Hello, The low-level API allows adding a C++ class via the add_class() method, and the high-level API requires using run() which does not support this feature. Could you please confirm for me that it is indeed impossible to have a multi-class python server that loads both HL python class and C++ class (shared lib)? And in case of impossibility, could you tell me if it is planned that this functionality be added? Thanks for your help, Arnaud

# 2 years ago

Hello,

The low-level API allows adding a C++ class via the add_class() method, and the high-level API requires using run() which does not support this feature.
Could you please confirm for me that it is indeed impossible to have a multi-class python server that loads both HL python class and C++ class (shared lib)?
And in case of impossibility, could you tell me if it is planned that this functionality be added?

Thanks for your help,
Arnaud

# 2 years ago
Anton	Hi Arnaud I think it might be possible to mix C++ classes with high-level API devices. You will have to use the tango.Util class directly. This is a singleton, so if we use add_class before calling tango.server.run, then it should see all the classes. The tango.server.run function is adding the Python classes internally. Based on the example here: https://pytango.readthedocs.io/en/v9.3.5/howto.html#multiple-device-classes-python-and-c-in-a-server (note: the example above has a typo on line 5 - it should be "util =", not "py =".) `import sys import tango from tango.server import run from IRMirror import IRMirror from PLC import PLC if __name__ == '__main__': util = tango.Util(sys.argv) util.add_class('SerialLine', 'SerialLine', language="c++") # run device server with SerialLine, PLC and IRMirror: run([PLC, IRMirror])` There is also a dict form that can be passed to run for more complicated settings, but this still doesn't allow a C++ class to be specified. It may be useful to extend the dict-form to suit your use case. Please let me know if the suggested code works or fails. /Anton

# 2 years ago

Anton

Hi Arnaud

I think it might be possible to mix C++ classes with high-level API devices. You will have to use the tango.Util class directly. This is a singleton, so if we use add_class before calling tango.server.run, then it should see all the classes. The tango.server.run function is adding the Python classes internally.

Based on the example here:
https://pytango.readthedocs.io/en/v9.3.5/howto.html#multiple-device-classes-python-and-c-in-a-server

(note: the example above has a typo on line 5 - it should be "util =", not "py =".)


import sys

import tango
from tango.server import run

from IRMirror import IRMirror
from PLC import PLC
if __name__ == '__main__':
    util = tango.Util(sys.argv)
    util.add_class('SerialLine', 'SerialLine', language="c++")

    # run device server with SerialLine, PLC and IRMirror:
    run([PLC, IRMirror])

There is also a dict form that can be passed to run for more complicated settings, but this still doesn't allow a C++ class to be specified. It may be useful to extend the dict-form to suit your use case.

Please let me know if the suggested code works or fails.

/Anton

# 2 years ago
alemeillour	Thank you very much Anton, it works fine with this syntax. I did a basic test in a Conda environment with pytango 9.3.3. import sys import tango from tango.server import run, Device class MyPowerSupply(Device): pass class OtherPowerSupply(Device): pass class MyServer(tango.Device_4Impl): pass class MyServerClass(tango.DeviceClass): pass #================================================================== if __name__ == '__main__': try: print(tango.__version__) util = tango.Util(sys.argv) util.add_class('Modbus', 'Modbus', language="c++") run([MyPowerSupply, OtherPowerSupply, [MyServerClass, MyServer]]) except tango.DevFailed as e: print('——-> Received a DevFailed exception: {}'.format(e)) except Exception as e: print('——-> An unforeseen exception occured…. {}'.format(e)) This allows great flexibility in the use of Python multiclasses Arnaud

# 2 years ago

alemeillour

Thank you very much Anton, it works fine with this syntax.
I did a basic test in a Conda environment with pytango 9.3.3.


import sys
import tango
from tango.server import run, Device

class MyPowerSupply(Device):
    pass

class OtherPowerSupply(Device):
    pass

class MyServer(tango.Device_4Impl):
    pass

class MyServerClass(tango.DeviceClass):
    pass

#==================================================================
if __name__ == '__main__':
    try:
        print(tango.__version__)
        util = tango.Util(sys.argv)
        util.add_class('Modbus', 'Modbus', language="c++")

        run([MyPowerSupply, OtherPowerSupply, [MyServerClass, MyServer]])

    except tango.DevFailed as e:
        print('——-> Received a DevFailed exception: {}'.format(e))

    except Exception as e:
        print('——-> An unforeseen exception occured…. {}'.format(e))

This allows great flexibility in the use of Python multiclasses smile

Arnaud

# 2 years ago
Anton	Thanks, Arnaud. That's good news. It would be great if you could propose a documentation update for PyTango, so others can find this in future. Fixing, and extending this section: https://gitlab.com/tango-controls/pytango/-/blob/v9.3.5/doc/howto.rst#L790-815 /Anton