Meeting Minutes

Meeting Minutes === :::spoiler Document Usage This document is used for all minutes of the WoT CG. After the meeting, the contents will be copied to the GitHub with a PR. === ###### tags: `Meetup` :::spoiler Document Usage This document is used for all minutes of the WoT CG. After the meeting, the contents will be copied to GitHub with a PR. For one week, changes can be requested and then the PR will be merged. After the PR is merged, the content here can be deleted. Document Access: - Everyone has read and commenting access - Cristiano Aguzzi and Ege Korkan can edit the document - Other people can request write access for a meeting - The scribe of the meeting will have write access   ::: :::info :date: **Date:** 18 November 2024 ### :bust_in_silhouette: Participants  - Ege Korkan - Cristiano Aguzzi ### :scroll: Agenda :writing_hand: **Scribe:** Ege :computer: ::: ---- ### Introduction Ege: Welcome to the 22nd meetup! ### News Cris: the week after we will have WoT week a physical event dedicated to WoT. Cris: We have office hours Cris: let us know if you want to present something! #### WoT Week Cris: We will see each other from 25-29 November in Munich. Two venues, Siemens and Microsoft. Cris: During the event we will host a Plugfest event an interoperability hackaton! Cris: Wednesday it is demo time, you can show your WoT solution. Cris: let us know if you want to join! ### Meetup Presentation V: I work at CALA. We accelerate lasers to measure ions. It is remotely controlled due to high power laser or radiation. V: You can measure the pulse or the cross section. Which allows you to get physical characterics. So there are data analysis steps as well. V: Data acqusition is IoT in the end. V: High power lasers by streching, reducing the frequency and then amplifying. In the end, it is compressed, where you get a high power pulse. V: All these steps use custom built components. The whole setup cannot be bought. > Vignesh shows the UI with different cameras and controls. V: Here we can use WoT to work with the integration time value in the UI. V: When we look at the devices, e.g. an oscilloscope, it has properties like time resolution. Actions like start acquisition. Events like measurement points. V: However, those devices also have state machines. When a measurement is happening, you cannot the trigger value of a camera. 1 out of 4 cameras can do that. > Vignesh shows the code part of Hololinked V: from the code, we can autogenerate the TD. > Vignesh shows property, action and event generation. V: We can add state information. You can tell if an action is allowed in a state or not. V: This is also reflected in the UI. V: One important thing is to correlate the timing of different measurements. V: Another part is how we can synchronize GUIs. Say, one user starts measurement from one PC. V: GUI information about the state is also very important and it is device specific. So we can notify the other clients about state changes. V: You can use the context API of frameworks like React or Svelte. V: That is it for today. I can get ### Question Cris: Is the last event implementation built on top of MQTT? V: It is on ZMQ. You don't have to define multiple socket. Messaging contract is different. V: We don't have a ZMQ binding though. It is interesting and the WebThing Protocol can be used by over ZMQ. V: Images need to be a separate frame though. Cris: What if you encode the image in JSON with base64 then you don't need another frame. V: I also could not get nice image manipulation in JS so I use the python qt for that. Richard: How do you handle big arrays like numpy? V: You can preencode it in zeromq. You can dump the memory on the message quite easily. Richard: Is there a standard way in msgpack (or JSON or whatever) to serialise a big numeric array? I'm wondering if there's something that's not numpy-specific? V: Python's own parser was much more performant than numpy. C++ and JSON Ege: How is the state infomration conveyed to the Consumer? V: there is a property but the machine itself is in my head. I hope that the standards can help. Cris: You can also show different TDs per state, as long as there aren't so many states V: I also do a lot of time-critical processing. Like 1ms time intervals are important. Richard: I'm also curious how "dynamic" a thing description is allowed to be - I've sometimes assumed I can parse it once when I set up my client then assume it won't change: is that a bad idea? Cris: there are some use cases to ask for dynamic TDs but they are not that developed yet. So you are doing correctly. Ege: What does it mean to delete a property btw? V: I want to free up the memory. I can either write an empty array or send a DELETE request. Ege: So it is more like resetting Cris: Or clearing. V: I also fire a property update when there is a write. Ege: How do you do async actions? V: You can add `synchronous:False` which starts a thread but you cannot do anything with that thread. Ege: Maybe you can combine the devices you are bringing with the openflexure microscope we have. Richard: It would cool to test if the UI can control the microscope backend. I also have a new implementation that is almost ready. You can use a new OS image. ## :ballot_box_with_check: Resolutions ## :exclamation: Action Items ## :envelope_with_arrow: Feedbacks - State machine modeling is missing. - Long running actions is done differently between two implementations (hololinked vs openflexure) ## Presentation Summary - Deleting properties? what is that - Async actions -> How to handle it later on? - For state machine, what is generated in the TD?