The conclusion I come to from this is that yes, I actually do need tmux, as the alternatives proposed are far more annoying and provide no benefit. I don't have a need for graphics in the terminal, and frankly I find it odd that we wouldn't simply display graphics... with the graphics system. But I do have a need for seamless session persistence and multiple terminals, and I do enjoy splitting a window when I'm running a command on multiple servers.
The Element app on Linux is currently broken (if you want to use encryption, so basically for everyone) due to an issue with Electron. Luckily it still works in a regular browser. I'm really baffled by how that can happen.
Only if the motor were in the hub of the wheel, which given the typical size of the hubs, seems even less likely. Remember that bicycle drivetrains are typically one-way due to the ratchet, so you can't apply braking force via the chain.
D-sub has got to be one of the longest enduring connector standards I can think of, apart from wall outlets. They're from the 50s, originally for military use, and we're still speccing them in new space hardware today. Now they've got coax/twinax, high power, fiber, and even pneumatic "contacts" if you know where to look (and can afford it). I can't say that they'd be my first choice, personally, but it's quite remarkable to see how well they've fared over the better part of a century.
XLR used (mostly) in audio is also from the 1950s.
The biggest problem with these standards is they are used for everything and so you cannot be sure that if the cable fits it will work. If a USB cable fits it will almost always work - but if it doesn't it will be obvious to your average idiot way (that is you can plug a mouse into a power supply - but nobody expects it will work). USB-C somewhat violates that, but even still it mostly is a case if you can get the connectors to fit it works.
When the plug is inserted, the jack "breaks its normal connection." Like they didn't want to leave the audio output like a floating pin to reduce stray voltage?
Scribner calls the switch "spring-jack" after "jack-knife" where the "jack" part of it comes from the name Jack and in the 1300s meant a mechanical device. So the "female" component of the connection was thereby given a "male" name.
Charles E. Scribner filed a patent in 1878 to facilitate switchboard
operation using his spring-jack switch. In it, a conductive lever pushed by a
spring is normally connected to one contact. But when a cable with a
conductive plug is inserted into a hole and makes contact with that lever,
the lever pivots and breaks its normal connection. The receptacle was called
a jack-knife because of its resemblance to a pocket clasp-knife. This is said
to be the origin of calling the receptacle a jack. Scribner filed a patent in
1880 which removes the lever and resembles the modern connector and made
improvements to switchboard design in subsequent patents filed in 1882.
late 14c., jakke "a mechanical device," from the masc. name Jack. The proper
name was used in Middle English for "any common fellow," and thereafter
extended to various appliances which do the work of common servants (1570s).
I don't think it was about not wanting to leave the audio input floating. Rather the "normal connection" is that the telephone subscriber is connected directly to the switchboard operator's annunciator (a display panel) so that the subscriber can light up a bulb on the annunciator when that subscriber wishes to ask the operator to reroute that subscriber's connection to another subscriber (instead of to the switchboard operator). This is why the switch ought to act like a double-throw, not just a single-throw switch. I think something along those lines is the reason...
> In a telephone-exchange system the wires of the several subscribers are run into a cen tral office, where, upon request, any wire may be connected with that of any other subscriber.
> In Fig. 4 is shown the cut-out connected with subscriber's wire in and the relay and annunciator P and O, and also, with the operator's telephone J, by means of the plug A, which is provided with a metallic point, and conducting-cord d. The connections are formed as follows: The subscriber S, by throwing on his local battery, sends a current along the wire in through the relay P, which, closing, the annunciator number of S is indicated at O, and the current passes along the Wire H, and thence through the switch to the ground Wire G.
Didn't even think of that, yes of course XLR and for that matter, 1/4" TS/TRS connectors were originally for switching phones at AT&T, before automated switching. Incidentally, you can also blow up quite a bit of stuff with them, depending on whether they are at consumer "line level", pro audio "line level", or even speaker level. We're definitely too comfortable with "if it fits, it works" (or at least isn't harmful".
There was a wild period in early transistor electronics where DC power adapters sometimes used 1/8" phone jacks - before the barrel-style DC plugs became common. Having 9V DC in a form that could be plugged into a microphone input always seemed like a pretty insane choice.
I'd like to mention my USB-stick-shaped audio recorder/player who's headphone jack (only uses built-in mic in any case, though iirc the headphone button skips to the next file) functions as the computer and charging connector.
The device was shipped with a cable (USB-A male) <> (TRRS 3.5mm aka 1/8") for this usage. It reports as mass storage.
Though that’s both older and better than RCA plugs. Somehow both RCA and NEMA managed to spread, the former world-wide, the latter only US + colonies, despite much better alternatives already existing and the drawbacks being obvious even 100 years ago.
I hypothesize that the car cigarette lighter port is the only truly standard, widely-used connector remaining. Wall outlets, just for 120V in the US, have at least three variants. 3.5mm audio plugs have at least four variants.
Don't speak in absolutes, if you like being correct. In an engineering context, it's quite important that you know which one is which. I can and have ordered a B size shell with 9 pins. This does not look particularly similar to whatever you are imagining - there are large power pins in the shell.
Because we codified the nomenclature, the difference is important, and the standard "serial port" is a DE9 and nothing else. The word "standard" wasn't codified, but D-sub connectors were.
D-subminiature connectors are codified by both IEC 60807-3 and MIL-DTL-24308K.
Neither IEC 60807-3 nor MIL-DTL-24308K "standardize" or "codify" D-subminiature connectors into DA/DB/DC/DE sizes.
Is there an actual standard referencing DA/DB/DC/DD/DE? It wasn't linked in the article.
I do not think there is, and I think that everyone claiming that DA/DB/DC/DD/DE is a "standard" is wrong.
After all, we 100% DEFINITELY want to be "correct". Words like "standard" have meaning.
It appears DA/DB/DC/DD/DE is just a trade practice started by Cannon. Maybe that's why the "standardized" and "codified" specifications refer to sizes 1 through 5 (or 6).
If we want the opinion of the ultimate arbiters of standardization, both Digikey and Mouser adhere to "the standard" by organizing shell sizes into IEC 60807-3 and MIL-DTL-24308K-compliant numerical sizes with letters in parentheses to denote that the letters ARE NOT a standard.
The most likely reason that DA-DE sizes are not in the standards is that DA-DE were once trademarks or otherwise proprietary designations created by Cannon. Indeed, practically the only consistent and quasi-official spec sheets that list the A-E sizes are published by ITT Cannon but even they reference the actual standards (e.g. "E Size 9 (MIL-DTL-24308 Size 1)").
I assert that DA-DE are proprietary designations created by Cannon (now ITT Cannon) and calling them a "standard" is incorrect, IN AN ENGINEERING CONTEXT.
In support of my position I have referenced both IEC 60807-3 and MIL-DTL-24308K and provided real-world examples from domain experts. I have also found pdfs for DIN 41652, CECC 75301-802 and referenced spec and marketing materials for Amphenol, Assmann, and Farnell/Newark and the only instances of a "standard" is when they list A-E sizes as an afterthought to aid people who are not following the actual standard to source standards-compliant parts (or ITT Cannon).
What is there, besides blog posts, to show that I am not correct?
edit: As a certified, triple-audited, ISO 9001-compliant weirdo, I am going to write up a nonconformity report, digitally sign it, print it out, manually sign it, then stamp it, then initial the stamp, then get it co-signed, stamped, and initialed, then scan it, then upload it into BMS, then print it out again, write the document control number on it, stamp and initial next to the document control number, have a second engineer stamp and initial it, and then hand it DIRECTLY to Quality if anyone ever refers to D-Sub connectors using non-standardized nomenclature ever again.
This is serious business and we are serious engineers here.
Besides the fact that you're clearly not taking this seriously, anyone can codify whatever they want, including Cannon. They invented them and wrote the spec, along with the nomenclature. If you want to use the names, use them correctly, otherwise use something else.
To which extent can a man go to not accept is wrong in an argument…
DB9 and RJ45. Period.
I have never once in 45 years of working with open system had somebody misunderstand what you mean with RJ45 or DB9.
That kind of pedantry can be saved for worst errors, like “baud-rate”. There is no such a thing. Baud is a unit [symbols/s]. It could be symbol-rate. I do have seen it bite a team, because symbolism rate was different as bit-rate. And still, I do not think is critical 99% of the time
I'd accept I was wrong if I was wrong, but I'm not. The standard 9-pin serial connector is a DE9 and nothing else. Cannon, the original manufacturer, says so, and no reputable sources say otherwise. Maybe you haven't been confused in 45 years. Other people could be. You could learn something and get it right, instead of being stubborn and objectively wrong.
Any significant benefits at 3 or 4 bit? I have access to twice that much VRAM and system RAM but of course that could potentially be better used for KV cache.
So dynamic quants like what I upload are not actually 4bit! It's a mixture of 4bit to 8bit with important layers being in higher precision! I wrote about our method here: https://docs.unsloth.ai/basics/unsloth-dynamic-2.0-ggufs
For coding you want more precision so the higher the quant the better.
But there is discussion if a smaller model in higher quant is better than a larger one in lower quant. Need to test for yourself with your use cases I'm afraid.
e: They did announce smaller variants will be released.
I can say that this really works great, I'm a heavy user of the unsloth dyanmic quants. I run DeepSeek v3/r1 in Q3, and ernie-300b and KimiK2 in Q3 too. Amazing performance. I run Qwen3-235b in both Q4 and Q8 and can barely tell the difference so much so that I just keep Q4 since it's twice as fast.
LLMs usually have about 3.6 bits of data per parameter. You're losing a lot of information if quantized to 2 bits. 4 bit quants are the sweet spot where there's not much quality loss.
I would say that three or four bit are likely to be significantly better. But that’s just from my previous experience with quants. Personally, I try not to use anything smaller than a Q4.
Is that really why it's going away? I just sort of figured that BLE chipsets had gotten so cheap these days that it was more economical. Can't see why you'd care about encrypted data for something which you could measure pretty trivially via other means, if you were close enough to pick up the signal.
Considering that if you DO use VAD (voice activity detection), it's the best open weights voice recognition model by a very wide margin, it's quite good. I'd be willing to be that commercial products that "don't have this problem" are using VAD as well, and that this is well known to them. But Whisper is just the weights, and I suppose a simple reference implementation, not a full product.
reply