I get it, wanting to do it right and understand what you are doing. I've worked as a metrologist, which has informed a lot of my interest in instrumentation, data collection, and data sharing.
At this point in life, it's a time thing for me: investing the time to get really good at coding and working with repos and containers is something I just don't have the bandwidth for. It's not a lack of interest, but the time for me to do those sorts of things easily was about 30 years ago, lol. So the AI provides not just a convenience but really the only realistic way for me to accomplish my goal in a reasonable time. I have all sorts of reservations about AI. It has its places, its use cases. For me, this is was an acceptable use case: no one's life should be in danger, and it provides consistent, reliable data. It's sloppy, but probably way more functional than anything I could have gotten going any time this year. It's a starting point. Hopefully it can be made better from there. On Thursday, May 21, 2026 at 11:06:34 AM UTC-4 Vince Skahan wrote: > Interesting but risky. Having done this kind of stuff for so long my > continued fear is minimally functional unsupportable garbage. It gave you a > working solution ‘as a starting point’ which is fine but ‘shipping the > prototype’ rarely ends well. > > And 100% of what it told you was fiction. You”ll not learn best practices > not spot security or privacy holes, nor learn what good even looks like. If > that works for you, cool. You scratched your itch, and thanks for the > github repo which is always appreciated, > > On Thursday, May 21, 2026 at 7:13:44 AM UTC-7 WeatheredScientist wrote: > >> I think it's interesting how far the LLM enabled me to go. Just setting >> up the configs was interesting, I saw it make a lot of the same errors I >> would have been likely to make (appending text strings instead of >> replacing, for instance). >> >> Ultimately, it's working, and I've shared it on github and docker. But >> it's fair to challenge what's up there to make it better, and make sure >> there isn't anything unnecessary or sloppy. If anyone is interested in >> helping, I'm open to learning and improving. >> >> On Thursday, May 21, 2026 at 7:38:37 AM UTC-4 Greg Troxel wrote: >> >>> WeatheredScientist <[email protected]> writes: >>> >>> > @Vince, great questions. Keeping in mind, this whole build was heavily >>> > AI-assisted, i used claude to answer some of your questions, but i >>> think >>> > the responses are reasonable. Please challenge if they are slop! >>> >>> Different people have different norms, but I think it's not ok to ask >>> other humans to wade through and fix LLM output. It's usually overly >>> verbose, and wrong in ways that are harder to spot than an actual >>> confused human. >>> >> -- You received this message because you are subscribed to the Google Groups "weewx-user" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/weewx-user/13972f8d-ac8c-49dc-8ef6-fc3757ff8301n%40googlegroups.com.
