
Continuing straight from yesterday: I had a freshly flashed node sitting on my bench with no configuration on it, and today's job was to actually make it useful using mesh.siliqs.net, the web-based Configurator. Same Web Serial/USB setup as flasher.siliqs.net yesterday, so at least that part wasn't new to me.
The Configurator walks through a handful of things that, until today, I only understood as concepts scattered across separate days of notes. Setting the node's role — field/sensor mode or gateway mode, the CLIENT versus CLIENT_MUTE distinction I wrote up yesterday — was the first choice, and it's the one that determines everything else about how the node behaves on the mesh. I set mine to field mode since that's what I had a bus to test against.
Then region and modem preset — which connects directly back to day 2's RF regulations notes and day 7's modulation parameters. The region setting is what keeps the node inside legal frequency and power limits for wherever it's actually deployed, and the modem preset is trading off range against data rate and airtime, the same tradeoff I wrote about weeks ago but had never actually had to pick a value for myself until today.
Then channel name and PSK — straight out of day 10. Typing in a channel name and a pre-shared key here is what actually puts this node on Siliqs' private mesh instead of some default public one, and doing it through the Configurator instead of raw serial commands meant I didn't have to worry about getting an exact command syntax right, just fill in a field.
Then the Modbus poll plan — which registers to poll on the local RS485 bus and how often, tying back to day 17's register types. This is the part that made the whole exercise click for me, because I could picture the alternative: without this tool, setting a poll plan means either editing firmware source and reflashing every time the poll list changes, or typing raw serial commands with a syntax you have to look up or remember exactly. Neither of those is something you'd want a field technician doing on-site. The Configurator turns it into filling out a form.
After setting everything, I read the config back — the tool has a way to pull the current settings off the node and display them, and I checked that what I'd set was actually what stuck. It matched, which was a relief, but it also made me appreciate why that read-back step matters: I don't want to assume a write succeeded just because the tool didn't show an error. Trusting the actual on-device state over my memory of what I clicked feels like a habit worth keeping generally, not just for this tool.
What I want to ask Wei about next: how much of what the Configurator writes is stored the same way whether you set it through the browser tool or through raw serial commands — I'm assuming it's the same underlying config store either way, with the Configurator just being a friendlier front end, but I haven't confirmed that and don't want to state it as fact yet.

直接接續昨天:我桌上有一台剛燒好韌體、還沒任何設定的 node,今天的任務是用 mesh.siliqs.net 這個 web-based Configurator 把它真的用起來。跟昨天 flasher.siliqs.net 一樣是 Web Serial/USB 的架構,至少這部分對我來說不是新東西。
Configurator 走過的幾件事,在今天之前,我都只是分散在不同天筆記裡的概念而已。設定 node 的角色(role)——field/sensor mode 還是 gateway mode,也就是我昨天寫的 CLIENT 跟 CLIENT_MUTE 的差別——是第一個要選的東西,也是決定這台 node 在 mesh 上其他所有行為的關鍵選擇。我設成 field mode,因為我手上有一條 bus 可以拿來測試。
接著是地區跟 modem preset——這直接接回 day 2 的 RF 法規筆記跟 day 7 的調變參數。地區設定是讓 node 待在實際部署地點合法的頻率跟功率範圍內,modem preset 則是在傳輸距離跟資料速率/airtime 之間做取捨,這正是我幾週前寫過的 tradeoff,但一直到今天才真的要自己選一個值出來。
再來是channel name 跟 PSK——直接來自 day 10。在這裡輸入 channel name 跟 pre-shared key,才是真正讓這台 node 加入 Siliqs 私有 mesh、而不是掛在某個預設公開 mesh 上的關鍵動作,透過 Configurator 做這件事,而不是打原始 serial 指令,代表我不用擔心指令語法打得對不對,只要把欄位填好就好。
然後是 Modbus poll plan——要 poll 本地 RS485 bus 上哪些 register、多久 poll 一次,接回 day 17 的 register 類型。這部分是讓整件事真正在我腦中「喀」一聲接上的關鍵,因為我可以想像沒有這個工具的話會是什麼樣子:要設定 poll plan,要嘛去改韌體原始碼然後每次 poll list 一改就重新燒錄,要嘛打原始 serial 指令,語法還得查或死記。這兩種都不是你會希望現場技術人員在客戶site上做的事。Configurator 把這整件事變成填一張表單。
全部設完之後,我把設定讀回來確認——這個工具可以把 node 上目前的設定值抓下來顯示出來,我對照了一下,確認我設的東西真的有生效。結果吻合,讓我鬆了一口氣,但也讓我體會到為什麼這個讀回確認的步驟很重要:我不該只因為工具沒跳出錯誤,就假設寫入一定成功了。相信裝置上的實際狀態、而不是相信自己記得點了什麼,這感覺是個值得養成的一般性習慣,不只適用在這個工具上。
接下來想問 Wei 的:不管是透過瀏覽器工具設定,還是打原始 serial 指令設定,Configurator 寫進去的東西,底層儲存方式是不是完全一樣——我猜底層的 config store 是同一個,Configurator 只是一個比較友善的前端,但我還沒確認過,暫時還不敢當成既定事實寫出來。