00:21:25	Andy Gotz:	@Anton you are right. There was a pre-PyTango version based on omnipy  (not based on the C++ library) but this was never really in production. Soleil took the decision to use the C++ library and wrap it with Boost.
00:23:56	Corne Lukken:	What is the default communication library used in PyTango omniORB or ZeroMQ, or are the both used in together?
00:24:10	Andy Gotz:	I think DESY (Teresa)?) worked on the Python Database server but this was a long time ago.
00:24:56	Andy Gotz:	omniORB for all synchronous and asynchronous calls and ZMQ for events
00:28:35	Corne Lukken:	Building a docker image in Gitlab CI requires the Docker in Docker (DinD) service if the gitlab-runner is using a Docker based executor. The Docker executor also must run in privileged mode. This still provides better isolation than using a shell based executor or forwarding the docker socket in the gitlab-runner config.toml. I have recently investigated this and be happy to provide information, an example I made lives here: https://gitlab.dantalion.nl/root/airglow/-/blob/master/.gitlab-ci.yml#L14
00:34:02	Benjamin Bertrand:	It’s possible to build docker images with gitlab-ci using kaniko: https://docs.gitlab.com/ee/ci/docker/using_kaniko.html
00:34:20	Benjamin Bertrand:	I did it recently for https://gitlab.com/tango-controls/docker/mysql/-/blob/main/.gitlab-ci.yml
00:39:39	Ugur Yilmaz:	it's also possible to use podman or other daemonless image builders
00:43:11	Igor Khokhriakov:	Hi everyone! Could someone please give more details why not to use pure python implementation? What was behind the decision to use cpp binding?

00:45:05	Andy Gotz:	@Geoff or @Anton will you talk about threading in PyTango ?
00:48:33	Corne Lukken:	Curse of the demo gods, happens to all of us
00:50:04	Corne Lukken:	Is the limitation of only being able to spawn a single device instance only applicable to PyTango unit tests or is this also a limitation when the library is used externally, I.E is it possible to launch multiple devices in the same memory space for integration / functional testing?
00:58:14	Andy Gotz:	@Come I am not sure to understand - we need an expert to answer.  You can have multiple devices in PyTango for sure. The C++ library is multi-threaded i.e. the omniORB layer is using multiple threads to listen to clients. Once you are back in Python you have to deal with the GIL. @Anton or @Geoff can answer your question better.
01:29:37	Samuel Twum:	@Anton thanks for the demo. Useful to see the tracing of functions as you interrogate the device from the proxy interface.
Qn: How is it that you're using the breakpoints without getting an error from the 3s timeout. I always run into that issue using the debugger.
01:30:40	Adriaan de Beer:	Will the slides be available from the recording on Youtube / somewhere else?
01:33:04	Andy Gotz:	ESRF uses green mode a lot for the new beamline control system
01:37:44	Reynald Bourtembourg:	Is the write_attr_hardware method called when writing an attribute?
I didn't see it in the debugging session.
Is it called only when using the pytango low level API?
01:42:54	Reynald Bourtembourg:	Here is gdb command to avoid the timeout annoying issue when debugging:

handle SIGPIPE nostop noprint pass
01:43:34	Samuel Twum:	Thanks @Reynald
01:43:54	Corne Lukken:	Can you switch out executors in PyTango for instance switch from greenlets / synchronous executors to OS managed threads?