sam1714

During the heights of COVID, the bargain store by me had boxes of trays of American Airlines' Biscoff. They were clearly optimized for weight: packaged in a super lightweight plastic tray with shrink wrap so thin it already had holes in it.

They apparently had enough getting stale to offload them to a liquidator.

Current Starlink direct-to-mobile is proprietary. It's kind of the opposite: 5G NR NTN so far has required new chipsets which haven't found their way into phones yet, whereas Starlink's LTE has been deployed with firmware changes to existing phones.

> something the store managers can figure out

That's like saying the store manager should manage the store website and networks instead of the chain's IT department.

No grocery store manager is going to be a licensed refrigeration technician. Ask them if they know about basic refrigeration concepts like superheat and subcooling, high side and low side pressures, compressor oils...

Further, a grocery store wants to optimize efficiency down to the percent, because that will save tens of millions of dollars over the fleet, and CO2 emissions. A store manager hardly has the capability.

>Ignore may be a tad strong. I am merely asking for hardware that is subject to my control to be within my control.

Which is how they got into this situation. These are industrial refrigeration systems subject to a huge list of variables (how many units, their sizes, and doors hooked up to a refrigeration circuit, open vs closed cases, ambient conditions, hours of store operation, what type of refrigerant, whether heat is going to a cooling tower or air heat exchanger). They use a ton of energy, and percent optimization leads to millions of dollars over a fleet. They are designed to let the user dial in every operating parameter, like defrost.

Compare this to my home fridge where all these parameters are fixed in the firmware because the hardware is 100% identical.

Industrial systems are necessarily different from mass-produced home appliances.

>Where people work, who have to empty a fridge before it can be remotely defrosted.

You don't empty a fridge with built-in defrost. Your home fridge defrosts itself every 12-24 hours and you don't have to do anything.

Normally, the temperature recovers in ~1 hour after a defrost. The attack is apparently chaining manual defrost commands one after another so that the interior gets way too hot.

If you want to start a new trend, you should do this with FIPS 140, which has to be recertified every year. "Apple crypto is no longer FIPS certified"

Even better, last year's 26 releases are still undergoing the process.

So the best solution would be Video CD. The original VCD worked like an audio CD: random access, time search, chapters, trick play (in forward).

Later on, they added menus, called Playback Control, but many players let you turn that off.

E-commerce is the important use case, particularly in goods where people care about appearance: clothing, cosmetics, food.

Things in these categories have colorful packaging with spot colors, metallic inks, coatings, as well as store displays with carefully designed lighting. The added cost is justified because it increases sales. It makes sense to carry that over online.

Similar to why there's a significant fraction of pro-mainland China people in Taiwan and Hong Kong.

Reasons include people who benefitted from the Soviet Union, often due to business and trade. Analogy would be Foxconn, a Taiwanese company which is dependent on manufacturing in China.

Nostalgia/rose-tinted glasses/rosy retrospection, Ostalgie in German.

Families that were ethnically from Russia, moved during the Soviet Union times and still have ties to Russia.

The former East Germany is way more pro-Russian than the former West Germany, showing the effects of these factors.

>I wonder if they considered using the fact that a roller coaster trajectory is closed loop

In terms of the rider, this isn't always the case. Often, the unload station is separate from the load station, basically pipelining. Also for theming; for example, Disney's Space Mountain has the unload and load zones on separate floors.

The paper did use the fact that velocity was 0 when they started and ended.

Chipset models only form a weak lower bound on crappiness.

You can certainly wreck connectivity with bad and old firmware. Similarly, housing design, PCB thermal isolation, not putting the sensor next to the MCU, measurement averaging settings, etc. all are going to affect accuracy even with the same sensor (particularly dynamic accuracy).

The Chevy Volt (and it's predecessor the GM 2-mode hybrid, and the Allison power-split hybrid) was a work of art. It is fundamentally a planetary gearbox, with 2 motor-generators.

In ICE mode, under 35 MPH, it worked as a series hybrid. Above 35, the engine mechanically connected to the wheels. One motor-gen extracted power and fed it to the other, acting as a CVT in parallel. This provided the ratio changes needed for the small engine.

It was way too complicated and expensive the two times they tried it (2008, then the Volt). It was successful and lives on to this day in its original application: city buses.

No, you're misreading it. Information returns in most cases is the W-2 you send out to your employees and file a copy with the IRS for their information. I've always said that, for employment tax purposes, you use the LLC's EIN.

The entire quote on the IRS's webpage: "For example, if a disregarded entity LLC that is owned by an individual is required to provide a Form W-9, Request for Taxpayer Identification Number (TIN) and Certification, the W-9 should provide the owner’s SSN or EIN, not the LLC’s EIN.

For certain Employment Tax and Excise Tax requirements discussed below, the EIN of the LLC must be used. An LLC will need an EIN if it has any employees or if it will be required to file any of the excise tax forms listed below."

The original statement was SSN hiding for invoices, that's directly covered in the W-9 example. You can (have to) hide your SSN on W-2s, but not the W-9s you give to customers.

No, it's not orthogonal. You don't get that benefit for an LLC. It's written on the W-9 form itself. Thus it's a reason why _not_ to have an LLC.

"If you are a sole proprietor and you have an EIN, you may enter either your SSN or EIN.

If you are a single-member LLC that is disregarded as an entity separate from its owner, enter the owner’s SSN (or EIN, if the owner has one). If the LLC is classified as a corporation or partnership, enter the entity’s EIN" [0]

[0] https://www.irs.gov/pub/irs-pdf/fw9.pdf

>The EIN is just to avoid putting your personal SSN on invoices.

In order to do that, your LLC has to elect to be taxed as an S-corp, which requires it to file its own return. Worse, certain states and cities will impose a corporation tax on it.

Otherwise, if you retain disregarded entity status, you have to use your personal SSN for invoices. The corporation EIN is then mostly for employment tax purposes.

(Not actual tax advice)

"For example, if a disregarded entity LLC that is owned by an individual is required to provide a Form W-9, Request for Taxpayer Identification Number (TIN) and Certification, the W-9 should provide the owner’s SSN or EIN, not the LLC’s EIN." [0]

[0] https://www.irs.gov/businesses/small-businesses-self-employe...

>why can't it show the full problem so you could take a picture right as it happens and send it to your dealer for instant diagnosis?

Fault codes are uploaded to the manufacturer on modern cars with telematics. On mine, the app will say you have an engine fault (or need an oil change or recall) and offer to schedule an appropriate appointment. If you call them, they often have it already in their portal under your VIN. The manufacturer also keeps track of these for regulatory purposes (triggering recalls).

A single fault code doesn't mean much. That's like saying I got a segfault, please fix my problem. You need a stack trace, registers, what the user was doing at the time, whether they can reproduce it, etc.

OBD2 stores a _freeze frame_, most of its readings at the time of the fault. You also need to look at other fault codes, pending codes, frequency/history of the fault to make a diagnosis. In any case, you can't just give one code; it's 5 pages of readings.

Ah I understand what you're saying. That was the only way to multitask on the Mac on System 5 and 6. Even with System 7, using the _hide others_ option was common, which is still there today.

Even on big screens, I like MacOS's two level cmd-` to cycle between windows in the same app and cmd-tab to cycle between apps, and App Expose (4 fingers down on the trackpad). I know what I want to find is in another browser window or spreadsheet window.

I think there's a difference between concept and the implementation. I see screens as simply compositing. Different regions have different resolutions and color depths and you use display hardware to combine these together. For modern personal computers, we got this in Windows Vista and OS X 10.0.

Circa 1985 hardware has significant limitations and it's leaking through abstractions both ways. A modern program supporting deep color writes to an array consisting of R16G16B16A16 values, next to a legacy program writing R8G8B8 and even a window with 10-bit 4:2:0 YUV HDR video. On the Amiga, the hardware leaks through by having to deal with separate bit planes and palettes.

Similarly, the 3D hardware in a modern GPU easily upscales a legacy application's surface 1.25x to handle the mid-DPI laptop screen I'm using right now. 1985 hardware can't do that, because you're fiddling with the DAC output, you can only have one screen/ViewPort scan line, no overlapping regions, and if you look carefully, blank lines between regions for the hardware to reconfigure.

In the 50s, when they started to use machines to sort mail, there were keyers at every facility that sorted mail. I believe there are still are still some local keyers on parcel sorters, where it's harder to "lift" the image of the address on a large box.

For letter encoding, they used to have 55 centers, in 2014 it became one, in Utah. A large part of it is OCR getting better and deciphering most handwritten addresses.

Another part of it is simple consolidation: it's trivial to send images over networks, and due to how mail has to be sorted together for dispatch, the demand tended to be peaky. Consolidation lets you exploit different timezones. (Slowing down the mail also gives a big benefit because you can spread the load over the whole day)

It's not as complicated as it seems from a technical perspective. Stamps have been UV-tagged since the 1960s as an anti-counterfeiting/automated detection measure. In the analog era, you shine UV light on a letter and if it it glows enough, it's got a stamp.

Modern processing machines (AFCS) capture both a visible and UV image of the mailpiece, so it's fundamentally changing the image processing software to have it look for the barcode on the UV side if it can't find one on visible. The downstream DBCS sorters already read secondary barcodes that are applied by the AFCS which happen to be in UV ink.

Further, USPS benefits from the same idea to hide a barcode for marketing (junk) mail. When combined with a handwriting-like typeface, it makes letters look personal and gets people to open them more. Same psychology where mailers apply a bulk-rate stamp instead of using a permit imprint.

It's a very Steve Jobs/Apple case where it sounds like a minor detail that requires crazy advances, but under the covers, it's very technically feasible, and importantly, can exploit economies of scale.

The use case definitely exists today:

  - 3D apps (games) render at a resolution less than the desktop for performance reasons (even in fullscreen mode to avoid switching the video signal)
  - Similarly, streaming sticks/TVs render the UI at 720-1080p and overlay on hardware decoded 4K video (very common in anything that's not a console, Apple TV, or NVIDIA Shield)
  - Handling non-high DPI apps on a high-DPI desktop, or handling systems with mixed DPI displays
  - Combining HDR and SDR content on the same desktop. Same with deep-color apps.
  - UI effects like the OS X genie effect, app thumbnails

Force majeure can be defined within contracts. So if the contract said that force majeure included changes in law or failure to receive permits, then it's force majeure. There can also be a separate "change in law" provision.

One can't say whether it's valid or not without reading the actual contract and looking at the state's case law.

My understanding is the lots of small cores are for applications where you have lots of simple, independent tasks that tend to be IO-bound: web and other front-end servers, databases, transaction processing. SMT is optimal here because you can execute other threads while waiting for IO.

I haven't been closely following it, but the Intel E-cores are definitely less wide and have less execution units. The FPU/Vector hardware is 128-bits wide instead of 256. If you ask it to do 256-bit SIMD, the front-end breaks it into 2 operations.

It was an issue of fundamentally differing directions.

In the mid-2000's, Sun decided to take SPARC towards designs with many small SMT cores. In the era of single-core processors, the UltraSPARC T1 had 8 cores x 4 threads per core. This was at the same time Intel released the Pentium 4 with hyper-threading, so it was an industry trend.

This of course works great for very specific applications, particularly considering efficiency, but is awful for others. Scientific computation was especially bad because the T1 had only one FPU for 8 cores.

Fujitsu's SPARC64 didn't go in this direction, and stayed with a conventional design (2 way SMT at most). Sun realized this and started to also sell the Fujitsu SPARC64 for customers who couldn't use the thread level parallelism, an arrangement that lasted until the end.

The idea of lots of slow cores is still a thing today: Intel's Sierra Forest Xeon is 144 E-cores.

Fujitsu is basically the Japanese IBM.

Here's the companion, "we promise we fixed it."

https://www.youtube.com/watch?v=WBbA9uKP-mA

The vastly preferable situation was to have redundant wiring that properly complied with regulations, which was retrofitted in an airworthiness directive (safety recall) after the accident.

They also required engine software that detected the sudden increase in speed associated with a shaft break and automatically cut fuel.

Like Boeing, there was organizational dysfunction leading to a direct safety issue.

During the development of the A380, the German and French sides were using different versions of the CATIA PLM software (i.e. CAD), which prevented people from looking at a single unified model of the plane. One concrete problem it caused was major problems with the wiring harnesses, which led to a 2 year delay in the program, and 3 early A380's which couldn't be sold. (Airbus kept one for test, the other two got sent to museums)

That manifested in this accident because the wing wiring was routed through the engine disk burst axes (a hazard which is explicitly accounted for) with insufficient redundancy. That created the dangerous situation where they couldn't turn off the engine after landing.

Philips Hue. I've had issues with color ones, the White + Ambience ones are fine. (Makes sense as they omit the color section)

Hue is basically the only bulbs not being sourced from the same couple of white label manufacturers. Philips/Signify's rapidly shrinking line of dumb bulbs are being sourced from there too.

Never buy filament style bulbs. They are universally unreliable because of poor thermals with the emitters.

Lighting color is cultural. Go to Asia and you'll find many places, commercial and residential, lit with 5500K, traditionally fluorescents.

I've heard a theory that it has to do with genetics and lower eye sensitivity.