Jake Hamby

YouTube / Google Translate translates idioms when it knows them. "apples to oranges" turns into "comparar peras con manzanas" in Spanish, which is "comparing pears with apples".

Another LLM translation mystery: YouTube thinks "Pete" is a swear word in Spanish, so it translates Pete Hegseth as "[_____] Hegseth" (same for "Pete" in the Amazon ad with Pete Davidson).

YouTube video by PlanetaJuan Así es la ciudad MÁS RACISTA de USA 🇺🇸

Son muy valientes por filmar en el pueblo más racista de Texas hablando español. Sin embargo, no se ve a nadie caminando por la calle. Todos están en sus coches o dentro de sus casas. youtu.be/PlN2qVW7xpM?...

YouTube video by La Huella Dentro del CÁRTEL de SINALOA: sicarios, laboratorios y rutas a EE. UU. | La Huella | Documental HD

YT's auto-captioning is hallucinating words now. It subtitled "drogas" (drugs) as "drocas", and it's repeatedly calling "El Guëro" (the blond) "el gerero", which isn't a word ("soldier" is "guerrero", which he's obviously not saying). I shouldn't be correcting YT, lol. youtu.be/6WLKOHY_Z10?...

Por segunda vez, me sorprende el opulencio, esta vez por los mausoleos de tres pisos con aire acondicionado por los cuerpos del narcotraficantes fallecidos. Algunas de mausoleos costaron más de $300,000. (@ 11:01)

I was not ready for the photos of the $205 million in cash stashed away in the mansion (@ 6:50).

That's 2 million $100 bills.

YouTube video by Nat Geo en Español El Chapo Guzmán | Nat Geo en Español

Olvidé el detalle más interesante sobre El Chapo: el túnel de 1,5 km que sus cómplices construyeron para su fuga de la prisión de máxima seguridad, en dos ocasiones. youtu.be/w1Ze8AlFSnM?...

YouTube video by VICE Inside Spain’s Fascism Fandom

So, why are you learning Spanish? Is it for work, for travel, for culture? Or are you a Dutch guy who really loves Franco and fascism? youtu.be/sqKSXPiGe7U?...

Interestingly, both Be, Inc. and the post-Acorn owners of RISC OS tried to remain relevant in a Windows/Mac duopoly world by pivoting to "Internet appliance" use cases. Remember how those were going to be big? "Network Computers" running Java? chrisacorns.computinghistory.org.uk/AfterAcorn.h...

"Combating the popular notion that BeOS and Be, Inc. failed mostly at the hands of MSFT, this article will examine other factors that contributed to Be's downfall including a small dev team, underfunding, product maturity, events in the wider industry, lack of third party devs and apps, and timing."

I did love Be's old slogan, “one processor per person is not enough”. They were at least 10 years ahead of what typical PC hardware supported. You could buy SMP PC's, Macs, and UNIX workstations in 1997, but they were expensive and required workstation OS's, e.g. UNIX or NT. macfolkloreradio.com/be/

The PowerPC 603 was a lower cost variant, not intended for SMP use cases, so the cache coherency with RAM doesn't use the full "MESI" cache invalidation protocol that everyone uses: en.wikipedia.org/wiki/MESI_pr...

Instead, if one CPU *reads* a cache line, it gets flushed in the other CPU's cache.

While the BeBox was very cool looking, with the dual columns of green LEDs on the front of the case showing the CPU load of each of the two cores (in BeOS), it's not a very high-performance system. Not the fault of the motherboard designer but of the available Motorola chipsets for the PPC 603.

I bought an original BeBox in 1996 and even worked as an intern at Be, Inc. in Menlo Park for 6 months in 1997 (before returning to LA to finish my CS degree). I still have it, in my mom's carriage house (basically a large tool shed), but it no longer has a VGA card and other required components.

Gemini says: "There were roughly 1,800 BeBox computers ever made: about 1,000 of the original 66 MHz model and around 800 of the later 133 MHz version, making them very rare, collectible machines." Source: www.beunited.org/faq/index.php

Apparently the NetBSD/evbsh5 port was for an evaluation board for the Hitachi SuperH SH-5, a 64-bit RISC CPU that didn't achieve any market success. www.bsdnewsletter.com/2002/10/News...

OTOH, NetBSD still supports the BeBox, however many thousands are still in use. wiki.netbsd.org/ports/bebox/

NetBSD almost never removes support for a platform, but they do when it no longer works and nobody wants to fix it. www.netbsd.org/ports/histor...

Besides NetBSD/acorn26, they've removed pc532, playstation2, and evbsh5 ports. The PC532 was a homebrew computer design. en.wikipedia.org/wiki/PC532

"NetBSD support for Acorn machines was eventually provided by the acorn32 port for Risc PC and A7000 family models, along with the acorn26 port for Archimedes, A-series and R-series models, thus bringing a more modern Unix variant to Acorn's original Unix workstations."

NetBSD/acorn26 was removed.

"In performance terms, Acorn's R140 compared unfavourably with other 1989 models such as the Sun SPARCstation 1 and Digital DECstation 3100, particularly with regard to FP performance, although such disadvantages could perhaps have been overlooked in an entry-level workstation costing around £4,000"

"RISC iX was either supplied preinstalled on new computer hardware or was installed onsite from a portable tape drive by Granada Microcare, who would take the installation tape away with them."

"Once installed a backup of the core operating system to three floppy disks was possible…"

If you can acquire an original Acorn Archimedes that's not too new (or too old) and has a SCSI card and hard drive, you can install Acorn's port of UNIX, "RISC iX". I bet you didn't know this existed… www.4corn.co.uk/articles/ris...

More technical info here: en.wikipedia.org/wiki/RISC_iX

One current RISC OS community project is retrofitting proper VFP support into the OS. The original calling standard had 16 different variants, depending on:

- 26-bit vs. 32-bit
- implicit/explicit stack limit checking
- how FP args are passed
- reentrant or not

www.riscosopen.org/wiki/documen...

Old BBC BASIC programs should still work, except when they include inline assembly code that assumes 26-bit mode and expects CPU flags to be preserved across subroutine calls. I experienced this playing around with type-in programs from Acorn User magazine. Fixing that code could be a fun game.

Today's ARM has a different FPU called VFP, for Vector FP, although the original vector math instructions were removed and replaced with the "NEON" SIMD extensions, so VFP is a bit of a misnomer. ARM Linux went through a calling convention transition from "soft" to "hard" FP (using VFP registers).

What is emulated in RISC OS is support for the obsolete original FP math coprocessor, called "FPA". It's easy to emulate coprocessor instructions because they trigger an instruction abort exception that can be handled. Only a few RiscPC's from the 1990s had a hardware FPA. Otherwise, it's emulated.

Unlike z/OS, which can run binaries compiled in the 1960s, or x86 CPUs, which can run binaries compiled in the 1980s, it's not possible for RISC OS on ARM to run programs compiled for the original 26-bit processor modes. They had to do a 32-bit cleanup of everything. heyrick.eu/assembler/32...

Actually, I'm not thinking about mainframes today but legacy ARM CPUs, which had 26-bit addressing for the reason that the designers thought it'd be fun to put the arithmetic flags (carry, negative, zero, signed overflow) inside the PC register so subroutine calls and returns would preserve them.

IBM, I don't think "bittage" is in the dictionary. But this page is a useful explanation for why IBM mainframes have three different addressing modes: 24-bit, 31-bit, and 64-bit. Why 31 and not 32 bits? They thought it'd be fun to use one bit as a flag to indicate size. www.ibm.com/docs/en/cics...

There's a simple explanation for this: ice cream is loaded with saturated fat and sugar but doesn't contain "seed oils" or the other culinary boogiemen MAHA irrationally fears.

someone should make a list of every company with maga leadership, from local small businesses to massive corporations so we can boycott them.

I vote with my dollars and don’t want to support fascists