| 11 Jan 2025 |
 @hexa:lossy.network | what commit? | 15:29:54 |
| @hexa:lossy.network | c4c07753b9eefa9798c7ea3b209dc2fec434fb8d migrated us to numpy 2 | 15:49:07 |
| @hexa:lossy.network | that was during the 2024.12.1 upgrade | 15:49:21 |
| @hexa:lossy.network | on 2024-12-09 | 15:49:28 |
| @hexa:lossy.network | d5fc8adbbe607a78fd933729570a8e6e7475dece migrated us to python313 | 15:49:59 |
| @hexa:lossy.network | around the 2025.1.0 release on 2024-01-05 | 15:50:15 |
| @hexa:lossy.network | so I'm not sure how you are before 2024-12-09 and after 2025-01-05 at the same time 🤔 | 15:50:35 |
| @hexa:lossy.network | * around the 2025.1.0 release on 2025-01-05 | 15:50:39 |
| @hexa:lossy.network | possibly some downstream components? | 15:50:58 |
|  oak 🏳️🌈♥️ changed their profile picture. | 16:45:34 |
| oak 🏳️🌈♥️ changed their profile picture. | 16:46:35 |
 uep | zha here | 23:25:05 |
| @hexa:lossy.network | z2m | 23:25:25 |
| 12 Jan 2025 |
|  strutztm joined the room. | 00:21:20 |
 @rick:matrix.ciphernetics.nl | My main reason to switch would be more device support. Had to wait for 2025.1 for better support for some devices and generally zigbee2mqtt seems more hacker-minded and hacker-friendly | 00:37:00 |
| @hexa:lossy.network | yeah, much more introspection | 00:37:34 |
| @hexa:lossy.network | * yeah, much more introspection into the zigbee network was the reason I chose z2m | 00:37:46 |
 teto | started with zha since that was the default ?! but quickly switched to z2m as it supported more of my devices (and in general it supports more from what I read). Has been pretty good so far. It took hours to draw a map of my 10 devices network but that must be an issue with the network rather than z2m, which is why I was tempted to test version 2 before reporting some feature requests | 01:29:11 |
| teto | hexa: please share your nabu case device "voice" experience onceyou've tried. I ponder getting one. | 01:29:56 |
| @hexa:lossy.network | teto: did you read this article yet? | 01:30:33 |
| @hexa:lossy.network | tinker isn't entirely clear on what his processing pipeline is, but the general picture is there | 01:35:26 |
| @hexa:lossy.network | the local sentence patterns preform really well, but you need to know them | 01:36:00 |
| @hexa:lossy.network | piping stuff into an LLM allows you to send off wild queries, that it will execute perfectly fine | 01:37:00 |
| @hexa:lossy.network | it also holds context between multiple triggers | 01:37:20 |
| @hexa:lossy.network | but it sometimes messes up simple things | 01:37:55 |
| @hexa:lossy.network | earlier today I asked it to make a light brighter, and it did that, but also did that for a light in the room that was off | 01:38:18 |
| @hexa:lossy.network | which was unexpected | 01:38:20 |
| @hexa:lossy.network | * earlier today I asked it to make a light brighter, and it did that, but also did that for another light in the room that was off | 01:38:31 |
| @hexa:lossy.network | the packaging was cute and no plastics at all | 01:39:08 |
| @hexa:lossy.network | the device feels very light, it has rubber feet in the corners, but since it is so lightweight the cable sometimes dominates its positioning here 😄 | 01:39:54 |
