> Yes, but I still prefer a nice tone. Like why should I change my manners just because it has no feelings? If anything the statistic predicts a friendlier answer when I say "please".
Like you I generally prefer polite phrasings over rude ones, and I'm certainly not going to tell you how you should communicate to LLM models, however it's interesting to note that being somewhat rude and to the point seems to have some objective advantages, see for example:
Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)
I would hate for this style of communication to leak through to daily communications with humans (e.g. colleagues) though, so even just for that it can make sense to hold on to a polite form of communication :-).
I know at least about the first paper. But I'm pretty sure that heavily depends on the model and I'm not sure how reproducible those studies were.
I for myself use the trick to ask the model, after I explained what it has to do, what it thinks about my proposal. Of course I have no scientific evidence but to me it feels as it prevents some misunderstandings and the model follows more correctly my intention.
The conceptual relationship to PlanetScale seems somewhat dubious to me (as mentioned in other comments) but the idea of implementing snapshots and branching below the database server level, in the storage layer using Ceph, is neat and potentially quite elegant! Kudos for that idea :-). I hope implementing it works out to the same elegant API envisioned in the initial concept. I could see this being very useful in a development workflow like the example (dev-db/feature-login).
I have to say that my experiences with running virtualized relational database servers on top of Ceph centralized storage have shown somewhat disappointing performance, so I think that could become a real challenge if performance is or becomes a goal. However, I've encountered multiple mentions that Ceph performance is supposed to increase significantly as the number of storage nodes increases (from a handful to a dozen to much larger clusters) and while I cannot corroborate this from personal experience, it stands to reason that "throwing more (hardware) resources at the problem" can make a big difference.
I'm interested in more details around ceph if you can share. What type of setup, what database, any fine-tuning, what type of workloads, dataset volume etc.
I'm asking because when I carried out the research on ceph, it seemed to me pretty solid and not necessarily, to my surprise, easily bottlenecked
Make sure those mushrooms are prepared well though because as I understand it, the human digestive system has a hard time breaking down chitin cell walls, so without proper preparation, the mushrooms may pass through mostly undigested.
The point about letting the mushrooms soak up sunlight to boost their vitamin D content is totally true and absolutely fascinating though!
> When commonly consumed mushroom species are exposed to a source of ultraviolet (UV) radiation, such as sunlight or a UV lamp, they can generate nutritionally relevant amounts of vitamin D. The most common form of vitamin D in mushrooms is D2, with lesser amounts of vitamins D3 and D4, while vitamin D3 is the most common form in animal foods. Although the levels of vitamin D2 in UV-exposed mushrooms may decrease with storage and cooking, if they are consumed before the ‘best-before’ date, vitamin D2 level is likely to remain above 10 μg/100 g fresh weight, which is higher than the level in most vitamin D-containing foods and similar to the daily requirement of vitamin D recommended internationally.
I've recently run into and explored research explaining that natural sunlight stimulates energy production in mitochondria, resulting in an energizing and healing effect completely separate from vitamin D production.
I feel bad for posting this without references so here's one reference I could find right now:
> This study shows that longer wavelengths in sunlight are transmitted through the body. When these are presented via NIR LEDs in a laboratory-controlled environment at much lower energies, they again have the ability to be transmitted through the body and also are associated with improved visual function independent of ocular exposure. Hence, body penetration by longer wavelengths impacts systemically. Longer wavelengths improve mitochondrial membrane potential and ATP production and improve function in a species conserved pattern that can impact on mobility, visual function and cognition, particularly in ageing. In short lived animals they can extend average life span.
> Taken together, these data provide evidence of the importance of the full spectrum of sunlight for human health. They also highlight the potential dangers of the restricted spectra found in white LED lighting in the modern built environment that lacks longer wavelengths and whose output is generally restricted to around 400–650 nm. The absence of longer wavelengths from LED light sources may have implications for public health that should be addressed.
Funny that I should run into this now... Just this past weekend I tried the Home Assistant backup/restore mechanism for the first time, and it failed miserably for me :-(.
First it took over an hour to create the backup, then I got a 4.42 GiB tar file, that of course failed to upload to the new Home Assistant install.
I investigated and found that the tarball was just a compressed copy of the complete installation directory of my Home Assistant setup, and that included multi gigabyte `.cache/pip` and `.cache/uv` directories :-s (my old Home Assistant install operates from a Python virtual environment that I created, and Home Assistant keeps nagging me that this installation method is deprecated, so I decided to migrate to HAOS in a VM).
When I deleted those directories the tarball was less than 200 MiB but the new HAOS VM still would not accept the upload. All I got was "500 Internal Server Error - Server got itself in trouble". And of course because HAOS is an "appliance" its kind of a black box so I couldn't find out how to get access to error logs with details :-(.
In the end I decided that the path of least resistance was to simply start from scratch based on the HAOS virtual machine and take some days/weeks to build up the new Home Assistant setup before it's mature enough to take over from the old Home Assistant setup (which is running on hardware that is close to failure).
If you are at all comfortable with Linux system administration, manually setting up one or a handful of KVM/qemu powered virtual machines is not actually that hard at all (in my experience). If you like a GUI to guide your initial steps, "virt-manager" is pretty okay. I've been running 3-5 virtual machines for several years now based on a pretty vanilla Ubuntu Linux install (Debian would work just fine as well).
Now I do like a challenge every now and then, so I'm currently setting up Proxmox to gain live migrations and high availability for virtual machines, because I've become quite dependent on all of these services in virtual machines actually running successfully :-) even in the face of eventual hardware failure (like what happened to me in the past months).
IMO Linux system administration, KVM/Qemu, Docker, and virtual machines, and third-party tools in general are not something that should be involved in smart light bulb/sensor/pump etc management.
Task for an RTOS or no OS IMO. Or a single executable that runs on any OS without config. Should be simple, fast, "just work".
Home Assistant supports an absolutely massive number of both manufacturer and community maintained integrations that are necessary for a truly universal all-in-one home automation setup without vendor lock-in.
Plus, for the full HAOS experience (as a “server”) running add-ons that are convenient one-click installed Docker-based packages for popular 3rd party tools used for home automation (but not developed by Open Home Foundation themselves) like Zigbee2Mqtt, Frigate (DVR for IP cams), EspHome etc so you can manage everything in one central location.
You could definitely flip light switches and read sensors with a 20kb executable. But you’d sacrifice the core value-add of HA serving as the single lynchpin connecting every smart device you own today plus whatever you may add in the future.
I started with a 100% Philips Hue setup that forced me to use their app, and eventually wanted to add some unsupported Zigbee devices that Google Home didn’t do a good job exposing which pushed me to explore Home Assistant.
Since then I’ve added (and removed) countless different protocols, proprietary cloud integrations for robovacs or air purifiers, ESP32 boards I built myself, web cams, TVs, etc over the years with the only unchanging constant being Home Assistant at the center linking it all together.
> You could definitely flip light switches and read sensors with a 20kb executable. But you’d sacrifice the core value-add of HA serving as the single lynchpin connecting every smart device you own today plus whatever you may add in the future
Very unclear what is the rationale for this claim.
It's great that Home Assistant supports "n absolutely massive number of both manufacturer and community maintained integrations" but that is quite irrelevant. The point is that running what you need to (mostly) flick switches and measure temperatures should not pull huge infrastructure, software requirements, and hardware capabilities. If it does then it does mean massive bloat.
Sickness caused by bacteria doesn't happen as soon as one bad bacteria (bacterium?) enters your body, a certain critical mass is usually required. This is very similar to the concept of "viral load" where a certain amount of viral genetic material needs to be exchanged before the viral infection can take hold.
The "beneficial bacteria" on your skin and in your gut make it harder for bad bacteria to take root in many different ways, one of them simply being they provide competition, "crowding out the bad guys".
Another way is that many, many, many types of antibiotics were originally discovered as metabolites produced by bacteria and fungi (examples include penicillin, streptomycin, chloramphenicol, and tetracycline).
And for completeness sake, milk kefir contains many Lactobacillus species that are also a natural part of the mammal microbiome (which makes sense when you think about it; Lactobacillus are named for consuming lactose, an ingredient of mammal milk).
American independent journalism seems to be dying (unfortunately) but I think in Europe there are several large news organizations reporting on things that matter in a relatively independent fashion, at least a lot more independent than what we see happening in the US (I'm thinking of e.g. The Guardian, Le Monde, I could also name a couple of Dutch news sources, but they would mean nothing to 95% of the readers here).
I'm really glad to read that the wrecking ball that is Trump & Musk's "reducing government costs" initiative in the US government didn't take out this very important foundation of _worldwide_ IT security. I'm thankful the CVE Board seems to have adequately prepared for this eventuality, hats off to them.
In some places in the US the use of sewage sludge to fertilize farm land is turning into a shit show of epic proportions (sorry couldn't resist ;-) due to severe PFAS contamination:
I have to say I'm kind of happy were not doing this in the Netherlands, because PFAS are a horrible class of chemicals that are causing all sorts of issues in nature as well as human health (obviously the two are intricately linked). I'm kind of ashamed to hear that the Netherlands are trying to get rid of their sewage in the UK though...
Edit: Upon reading the investigatemidwest.org article I see that it's about the exact same subject, apologies if my reply is perceived as noise.
Like you I generally prefer polite phrasings over rude ones, and I'm certainly not going to tell you how you should communicate to LLM models, however it's interesting to note that being somewhat rude and to the point seems to have some objective advantages, see for example:
Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)
https://arxiv.org/abs/2510.04950
Understanding Tone-Dependent Inference Cost in Large Language Models
https://arxiv.org/abs/2607.23915
I would hate for this style of communication to leak through to daily communications with humans (e.g. colleagues) though, so even just for that it can make sense to hold on to a polite form of communication :-).