robryk @robryk@qoto.org

@8petros I assume the situation you are considering is a person without a fixed structure around them, who's wearing all the wind/water protection and insulation they are using. I also assume you want something that will not require constant attention and so allow the person to sleep.

I'm curious what sort of container are you considering. (I initially thought that it's very important that it be well sealed, but I realized that as long as no solid calcium chloride leaks out, the only potential danger from the contents I can see it its high temperature.)

@8petros

I expect that optimality would depend enough on what you care about: highest total heat delivered, highest rate of heat delivery over first XXminutes, highest heat delivery while keeping the heat delivery rate uniform, ...

Now that I think of it, if you want stable rate of heat delivery you might wish to feed water at a constant rate into a container with cacl2 instead of mixing it up at once.

@8petros What are you optimizing for?

IIUC adding too much CaCl2 will not hurt: it will just not dissolve once you go past the amount a saturated solution can take. So, if CaCl2 is cheap enough you can just add too much: the only downside is that your heatpack will have higher heat capacity (and so will use up more heat to heat itself up). Note that its solubility in water increases with temperate in a significant way (see e.g. the sidebar on https://en.wikipedia.org/wiki/Calcium_chloride).

I don't know exactly how quickly, but dissolution slows down when you get close to saturated solution. It implies that you might wish to be far from saturated solution so as to get a faster-acting heatpack. However, I expect that this effect can be ignored, because having more water means having more mass for the same total energy, so heating the heatpack up will become a larger factor in reducing the power provided to the outside.

tl;dr I would aim for around saturated solution at heatpack's intended operating temperature, unless you are optimizing for some specific unstated variables.

@kityates Unless they're operating around the clock, they _sometimes_ experience ~zero calls, so that might technically be true. :P

@capybara @kityates

Note that GP was talking about an "average" that's larger than _all_ the values being averaged, not most.

@matthew_d_green

Uh, either there was a typo along the way, or it's nowhere close to 2^16*10^6: 2^16 is 65536 as opposed to 65356.

@rysiek @matthew_d_green

It doesn't seem to be printed value of (any IEEE) float of 1M*2^32 either...

@dianor At least some of them are also value judgements on societies: while it's probably true that a society where people are more often on time is nicer to live in (ignoring the effort one has to put it to uphold one's part) and more efficient, a society where people never interrupt each other might or might not be nicer to live in or more efficient than a society where the expected amount of being interrupted is much higher.

@zero323 @rysiek @ocdtrekkie

> Interestingly enough, it appears that there is no straightforward way to opt-out anymore.

Wouldn't switching the UPC-provided AP to modem/bridge mode do that? (Not that it's reasonable to expect people to do that just to opt out of this feature.)

@rysiek The networking approach also helps with a different problem: things becoming obsolete due to the producer-operated infrastructure turning down. If something can operate without external communications, it doesn't have a hard dependency on such infrastructure.

@rysiek

The first thing would probably backfire in interesting ways, given things like washing machines that have an advertised "smart" feature that they use less water by weighing the laundry. (It would either get bundled with the not-beneficial-for-consumer smarts or would lead to interpretation woes around "well, but _this is exactly the feature_ that causes the water usage metric to be lower.)

@rysiek @ocdtrekkie

https://www.upc.ch/en/support/internet/wi-free/

(to see that it's about using people's APs, look at the "will it affect my internet connection" question, to see that it's available-for-those-who-make-their-APs-available, look at the "how do I disable it" question)

@rysiek @ocdtrekkie

UPC does something like that in Switzerland too. They allow the customer to enable/disable the feature: the incentive is that if you enable it, _you_ can use that wireless network on others' routers (there's no noticeable latency in flipping the setting, but I guess people usually don't want to bother flipping it back and forth whenever they travel).

@rysiek What do you mean exactly by smart functionality? Anything that involves (bidirectional?) conversation with an external service? (For example, is a VCRs purely local ability to pause recording over commercials a smart functionality?)

@mjg59

I don't know the answer, but I wonder whether you want presence of anyone at the physical location of the machine, or whether you care about it being some particular user being present there or someplace else. (Relatedly, do you want the condition to be unsatisfiable if the request is made by a user who has no local session?)

@b0rk Why does it wish to log in to a Mastodon instance? When I read the description of what your webtool does, I assumed (apparently incorrectly) that it's fetching ActivityPub thread by following the replies collections downward and inReplyTo links upward.

@jfbastien

One slightly abstruse disadvantage: currently it's possible to tell that someone is incorrectly depending on uninitialized values (at least at runtime, and often statically), because the condition can be evaluated. In the proposed world there is no automatic way of testing whether one is accidentally using the initial value of a variable without intending to, because there's no automatically detectable difference between "actually wanted to read the default zero" and "didn't want to read anything but a value that should have been written in there, but the value was accidentally not written yet".

So, we will make most of those CVEs harmless, but some fraction will remain as security issues, but won't be automatically findable anymore. Still, this would probably be a net benefit.

@gsnedders @jfbastien

How does it make sense? I fail to see how this is not simply a very weird case of missed optimization, given that the compiler was free to compile the code into something that zero-initializes stack variables earlier anyway (because any use of them would be undefined behaviour).

@lauren When I have just graduated my high school it was just getting WLAN coverage for the whole building. IIUC at least initially youtube was blocked on that network simply to conserve bandwidth (not of the uplink, but of the AP-to-devices link). Maybe that's the reason for blocking TikTok (which would not apply to chatgpt and like)?

@grrrr_shark

I mean, I was explicitly lazy, so please don't fault yourself.

To be frank, I'm somewhat miffed by the epistemics of many fedi posts (or maybe by combination of epistemics and presentation). I often see posts that provide some terrible evidence for a claim (e.g. that has a very obvious unaccounted-for confounder, or that fails to find an effect, but lack of effect is the conclusion taken from it, etc.) but are worded as if that is nearly incontrovertible evidence. Their prevalence makes fedi a non-time-effective way to try to figure out anything about the topic, and their presentation makes it uncomfortable for me to try talking with the author about them.

All that said, I feel like complaining about that is tantamount to trying to impose rules of discourse on others that are (a) not standard (b) probably not upheld well by myself quite sometimes. So, I'm completely unsure what would be a generally useful course of action other than trying to do better myself. Alas, I'm not very good at that in particular on some topics.