brigade

App store apps are required to run in a sandbox that denies file accesses without user permission.

Muse is not available in the macOS app store. Almost certainly because of the sandboxing requirements.

Thumbnailing happens in a background process and shouldn’t block the main thread; rather hangs are usually because fetching xattr over SMB is slow and Finder will block until it’s enumerated the metadata of every file in a folder.

Preview actually does use progressive rendering for large enough images, even from SSDs. You can tell because opening an image will be blurry for several seconds, even when the full size image could have been decoded in a quarter second or less.

You're wrong. AVIF does support spatial layering where the full-resolution image is derived from the lower-resolution layer, and cannot be decoded independently.

Yes, the format is not inherently progressive. But having the fine steps in progressive quality that JPEG-2000 and JPEG-XL have has less general usefulness than the amount of words complaining about it.

For like a decade, the "opt-in" to ACR was included in the ToS agreement needed to use any smart features. Then at least some brands made the opt-in separate, but made it look like a recommendation engine or "help us improve your experience" agreement.

LG adopted the "Maybe Later" dark pattern for opt-in to ACR, at least in some markets. Though I guess you never know for sure with A/B testing.

All 10 cores support the same extensions. It’s silly to implement an SoC any other way, and it’s been a decade since someone has last made that mistake.

Well the original article is also about RISC-V isn't a great ISA for embedded either, and how that's supremely disappointing because there's no inherent reason why the designers couldn't have learned from decades of ISA research and made it suck even a little less. OP doesn't even defend the ISA committee (unlike HN, somehow...)

As much as people complain about warts in other ISAs like x86, their design decisions made sense back when they were made (or at least, no one really knew better)

There's an order of magnitude difference in speed requirements between file format parsing and image decoding, then another order of magnitude difference to video decoding. Even rav1d reuses dav1d's assembly (most of the actual runtime) to approach its speed.

Still true for extended attributes, which Finder and Spotlight love to query.

Even backwards compatibility is optional in USB4. There are USB4 devices (SSDs at least) that will not function when connected to USB 3 ports.

Zero micro ops to be precise, that’s handled entirely at the register rename stage with no data movement.

Cortex A8 vsub reads the second source register a cycle earlier than veor, so that can add one cycle latency

Not scalar, but still sub vs xor. Though you’d use vmov immediate for zeroing anyway.

Fine, you made me look it up. M1 Air defaults to 2880x1800 on a 2560x1600 display, which is a 1.125x scaling. Nowhere near the 1.5x you claim.

Apple never supersamples by 2x. Default MacBook Air scaling was around 1.1x iirc.

OP dances around the key context that this isn’t hidpi, but rather a 3rd party hack that uses hidpi rendering to supersample their “native” 4k resolution by 2x, since the end result looks more pleasing to them than the native 4k render.

And that itself is a recent policy change from just in the last two months; as of January United's official policy [1] matched the FAA's in only requiring checked devices to be powered down

[1] https://web.archive.org/web/20260129152627/https://www.unite...

It’s really a question of whether a team believes bugs are defects that deserve to be fixed, or annoyances that get in the way of shipping features. And all too often, KPIs and promotions are tied to the features, not the bugs.

Plus, I’ve been in jobs where fixing bugs ends up being implicitly discouraged; if you fix a bug then it invites questions from above for why the bug existed, whether the fix could cause another bug, how another regression will be prevented and so on. But simply ignoring bug reports never triggered attention.

That's why I said generally - once testimony is compelled, it can no longer be used against you. And the definite exception for compelling your name is if the government already believes that you committed a crime and is trying to figure out who you are, and you cannot articulate specifically why your name could be incriminating.

5th amendment protections can include questions of identity, if the question of identity is relevant for incrimination. Like, if the government has a warrant for "Joe Smith", you're not required to testify whether that's you. It's usually a waste of time since could just prove it with the non-testimonial evidence that lead to your arrest, but the protection does exist.

First, you have the right to say nothing at all; there is no requirement to incriminate someone else to protect yourself.

Second, you can still generally invoke the 5th amendment during testimony even if you already claimed someone else did it. You aren't under oath until said testimony, so it still protects against you having to choose between committing perjury or self-incrimination, and doing so cannot be used as evidence of either.

Most of the world also doesn't have the same degree of protections against self-incrimination that the 5th amendment provides. If someone shot a person with my gun, while the police can obviously ask questions, in the US I have the right to not answer and force them to prove beyond a reasonable doubt who fired it.

It’s not a hard limit, especially if you aren’t pushing the frequency wall like Intel. AMD used to use a 2-way 64kb L1, Intel has an 8-way 64kb L1i on Gracemont, and more to the point, high-end ARM Cortex has had 4-way 64kb L1 caches since before they even supported 16kb pages.

ARM favored wider ILP and mostly symmetric ALUs, while x86 favored wider and asymmetric ALUs

Most high-end ARM cores were 4x128b FMA, and Cortex-X925 goes to 6x128b FMA. Contrast that to Intel that was 2x256b FMA for the longest, then 2x512b FMA, with another 1-2 pipelines that can't do FMA.

But ultimately, 4x128b ≈ 2x256b, and 2x256b < 6x128b < 2x512b in throughput. Permute is a different factor though, if your algorithm cares about it.

Your first example is a CPU limitation that Instruments doesn't model (does perf?), but is still mostly better than Intel chips that are limited to 4 dynamic counters (I think still? At least that's what I see in the Alder Lake's Golden Cove perfmon files...)

Your second example, is the complaint that Instruments doesn't have flamegraph visualization? That was true a decade ago when it was written, and is not true today. Or that Instrument's trace file format isn't documented?

A quick google suggests that British Columbia's building code only requires STC 50 which is "you can hear but not understand a neighbor's loud conversation" levels of isolation. Though maybe your city has stricter requirements?

STC 50 is a common requirement in the US too.

AFAIK Jyrki came after WebP was already announced to add lossless support; rather I’d consider Skal the creator inasmuch as it was originally just an image container for VP8 intra. He was working on WebP2 at the time Google rejected JPEG-XL and also was not involved in that decision.

Are there even any billboards in Cupertino?

As someone who did predominately use stack overflow through Google search… I remember that half the time the top result was someone asking the question I had, only for it to be duped to a different question that didn’t answer the original. So they failed there.

For people in general these data brokers are a primary source of information for spammers, both political and semi-targeted. So they share responsibility for making calls from unknown numbers useless.

gAMA dates back to the days when hardly anyone had a clue about color primaries and colorspaces, let alone commonly tried to synchronize them across displays. It's explicitly ignored by everyone if the file uses any more modern method of signaling colorspace information, so really OP needs actually to write a colorspace to the file (cICP, iCCP, even an sRGB chunk) instead of merely claiming on their blog they wrote a colorspace.

That would require that they understand the protocol stack they're using to send H.264 frames

Proper rate control for such realtime streaming would also lower framerate and/or resolution to maintain the best quality and latency they can over dynamic network conditions and however little bandwidth they have. The fundamental issue is that they don't have this control loop at all, and are badly simulating it by polling JPEGs.