Hacker News Reader

Comment by gumby | Hacker News Reader

The negative impact of Postel's suggestion ("be conservative in what you send, be liberal in what you accept", AKA the "Robustness principle") is worse on end users than developers in cases like USB.

USB is silently super backwards compatible. This is really good in a pinch: I need to get the data off this drive and even if it takes six hours because I only have a USB 2 cable I'm better off than if I couldn't do it at all.

The problem is the opposite case for users: They have a fast USB-4 drive (potentially 40 Gbps) and a USB 4 laptop but pick up a USB-2 only cable. Hard for most people to diagnise, especially as such a cable won't have an e-marker so the laptop can't even query it. Or a slow charging cable for their device, so they fancy laptop doesn't visibly charge, or only charges overnight.

Even for me these cases are hard to diagnose. And harder to understand than, say, using an inferior cable with your monitor so you et no signal at all.

View on HN

I recently threw out all of my USB-C cables that are longer than a foot but can’t charge at 100w, because I decided USB-C is a primarily a charging protocol, and I’ll use wireless, short cables, or Ethernet for data transfer and HDMI for display.

The matrix of charging + data speeds in USB-C is a nightmare and always resulted in having the wrong cable for the desired devices and usage.

You could also just buy higher quality cables, right? The problem is usually that most consumers see two USB-C cables and one is $10 and the other is $50 so they buy the $10 one, only later realizing (if at all) that the $10 only does data, or only does 50W, or w/e.

But if you actually pay attention it seems like cables work as advertised with reasonable consistency, at least for me.

They may work as advertised, but the problem is that the cables themselves have no indication of what was advertised in the first place.

The ability to know what the cable you’re holding can do is just not there. Unless you have some way of ensuring that all your cables are the same, you need to:

- discard/donate every cable that you can’t identify (as @brookst did) by its eMarker (though they’re required by spec) unless you own a specialized USB-C voltmeter or multimeter

- manually label every cable with all its unique specifications, either after running it through an eMarker reader or immediately after taking it out of the box

> the problem is that the cables themselves have no indication of what was advertised in the first place

Fortunately, this is changing, though it's shameful it took this long to get standardized, straightforward labeling: https://tidbits.com/2022/09/29/usb-simplifies-branding/

Just plug it in. If you can't judge the the speed by "feel", then run measure how long it takes to charge / transfer a file.

Not sure why you're being downvoted. Lots of phones show how fast they are charging (charging slowly, fast etc)

It's an expensive solution, but most Thunderbolt 4 cables have markings and they support pretty much everything (Thunderbolt, USB, Alt modes, high-wattage charging, etc.).

Because Intel has a certification program for Thunderbolt that makes many things mandatory that are optional (for no good reason) in the USB-C spec: https://news.ycombinator.com/item?id=28744448

> optional (for no good reason)

The reason is that the USB-IF is owned/ run by the manufacturers and many of them want to be able to make cheaper gear (TBF mandatory high tolerances was part of the “problem” with the original Thunderbolt too, lo those many years ago).

This is also why USB-A is not symmetrical as was originally designed: to save money in the connector.

/Off topic

I was going to ask how to test these things as I need to do some housekeeping too. I just found out about (and bought) a digital USB volt/amp meters. Here's an example. [0]

[0] https://www.amazon.com/Eversame-Multimeter-Voltmeter-Indicat...

edit:

I just read this write-up [1], which clearly explained why the 65W USB C charging hub I bought doesn't necessarily enable "super fast charging" on my Samsung phone. Chargers and cables need to meet PD3.0 standard with PPS support to dynamically negotiate the volts and amps to maximize power throughput safely. Otherwise, you get auto-downgraded due to sticky default voltages.

[1] https://www.gsmarena.com/dont_just_buy_any_45w_pd_charger_fo...

Curious if this device you linked would also work to test things like the power bricks themselves and not just the cables? Would love to get rid of some low performing cables as well as the older power bricks.

Edit: Based on the description it does sound possible.

I use a FNIRSI-FNB58 tester for this. It can read and display e-markers.

https://www.aliexpress.com/item/1005004665244812.html

I've been thinking I should have something like this. Is this "the best there is" in this price category, or is there an even better product that someone here wants to recommend?

> Is this "the best there is" in this price category

That was my conclusion when I looked into this about a month ago, especially if USB-C is the main focus. I remember a reviewer pointing out some deficiencies on the USB-A front (compared to other, simpler, meters) but unfortunately I don't remember what they were.

I was thinking of getting something like this, but wouldn't it get outdated as soon as there's a new standard, i.e. very fast?

That's definitely a risk.

The firmware on the unit is upgradable. While there's no guarantee that the manufacturer will (or can) add support for any given future standard, looking at the firmware changelog suggests they have at least some track record of doing this:

V0.65 2022-10-27 Add automatic detection QC4/QC4+/QC5

V0.63 2022-10-16 Compatible with other Xiaomi charging heads

V0.60 2022-10-07 Added support for PD3.1 triggering

V0.60 2022-10-07 Added support for Xiaomi A port PD charging head triggering

Sounds good. Thanks for the update.

Most people who want 40Gbps can diagnose the issues, the rest will assume "Oh, it takes 6 hours, better plan for it, that's how it is". But it will work every. single. time. at those very slow 2.0 speeds.

What they should do is give a popup saying "You have a slow cable, it could be faster, go rummage till this message goes away and mark the good one with a sharpie", and that wouldn't need any changes to the spec

They've chosen reliability over quality. Cheap reliability is a great thing. We did fine before fast hard drives, we can often live without them. But we'll want to scream if we can't get it working at all.

> go rummage till this message goes away and mark the good one with a sharpie

I dunno if I even have the right kind, or how to even shop for the right kind, because there are like 500 wrong kinds that are all wrong for a different reason =/

Basically I just look for Thunderbolt 3/4 support and if it doesn't have that, I give up, because who the heck has time to understand 4000 kinds of incompatible USB?

Maybe they could have a link in the popup?

I wonder if you could start an online cables boutique that just sold tested and reviewed cables and cable management with a 50% markup for people who just don't want to deal with it.

Then you have to wonder about how to make sure the link doesn't 404.

And your store idea invalidates having cheaper devices and cables and a ubiquitous protocol.

The only good thing about USBC is that if you moderately keep up to date with all your devices, eventually, recharging becomes relatively easy for all your devices. I don't think I've ever had a problem charging a USBC device. Generally everything charges everything. The high power devices like laptops are probably the exception since you need a brick that can deliver 65-80W[1].

1. One reason why Apple Silicon is so appealing. The M2 Air is plenty fast and basically consumes the power of an iPad. You don't need a 30W supply to charge it.

You're lucky. I've had a LOT of issues charging USB-C devices, like:

- devices that don't support USB Power Delivery

- chargers that don't put out sufficient power

- chargers that don't put out sufficient power only when there's more than 1 device plugged in

- devices that only work with a special USB Type A charger (the fat kind), not with C to C charging

- Google/Samsung phones not charging with each other's chargers, even though both have similar power draws and both use USB-C

- devices where only certain USB-C ports can reliably charge (some recent Macbooks had this issue)

- devices where the USB-C port can't be used for charging at all, only comms

- USB-C hubs with unreliable power supplies that sometimes work sometimes don't and eventually dies on you altogether

- USB-C cables that wear out unpredictably over time and only support slow charging after a while, not PD fast charging

etc. All of those were scenarios I've personally experienced in the last few years. It's a nightmare.

Who's "they"? The operating system vendor? I don't think there is even a single go-to place/brand to buy certified USB cables. The entire ecosystem is a mess, and even ordering first-party from Microsoft/Dell/etc is no guarantee that you'll get the right kind of cable, and I doubt that any of their support staff knows the ins and outs. I don't think even the standards body members themselves can figure that out at home.

Thunderbolt is much cleaner but much more expensive.

50% ıs too low. Anyone doıng this should aim for 5000% and also shout everywhere how their copper is so much better than anyone else's that it's worth many times its weight in gold.

If you just buy Apple's most expensive Thunderbolt/USB-C cable, it'll solve basically all your problems.

Except for the fact that it's very expensive.

I love having a ten-foot-long Thunderbolt 4 cable!

(So much so that I bought two of them.)

Same. Expensive but so worth it. I also have a 10 foot USB-C cable from Facebook I use with my Oculus

Just don't let it touch the floor at any point. /AudiophileAdviceMeme

That might work as long as the cable delivers its promise of a balanced soundstage and fast punchy mids!

Not good enough unless it also gives me sparkly highs!

You have to be sure to plug it in the correct orientation for that

> I dunno if I even have the right kind, or how to even shop for the right kind

I don’t know if you can trust what you see on eg Amazon but “USB IF certified” should be a fully-spec-compliant cable. I’ve successfully bought cables on Amazon this way.

It’s annoying, but from research that’s the magic incantation to search for.

> What they should do is give a popup saying "You have a slow cable, it could be faster, go rummage till this message goes away and mark the good one with a sharpie", and that wouldn't need any changes to the spec

I think something like this exists in Windows. I'm not sure under what conditions it triggers, but I remember occasionally seeing a tooltip saying something like "this device could work faster", directing me to plug it into a different USB port.

I remember this as well, but wasn't this way back if a usb 2.0 device was on a 1.1 (!) port?

Not sure if they continued this for USB 3.0 and USB-C.

https://devblogs.microsoft.com/oldnewthing/20070821-01/?p=25...

They did continue this for USB-C. My new laptop with USB-C charging port constantly whines that my new, after-market charger should be replaced by a $LAPTOP_MAKER charger to charge faster. The cable is theirs but the power supply is a multi-output charger instead of their original charger. At the moment I believe the problem is some sort of signaling rather than the capacity of the power supply.

I haven't seen that message in a long time, but I remember it being quite unhelpful as often there was no way to plug the device in to a port that made Windows happy. And no way to tell Windows you didn't care so stop complaining about stuff.

The only time I recall ever seeing this message myself was when I was writing USB device firmware from scratch. In my situation, the actual problem was that I had made a mistake in one of the Windows-specific device descriptors. So it’s possible that the device itself was wrong…

> What they should do is give a popup saying "You have a slow cable, it could be faster, go rummage till this message goes away and mark the good one with a sharpie"

Android does this. "Charging slowly"

You need to find mighty shitty cable and they aren't talking about charge speed, but transfer speed...

I have a USB-C cable thay came with my monitor which doubles as video and power pass through. If I use thay cable on a Dell USB-C hub, my Pixel reports it can only charge slowly. Cable doesn't matter.

There is no way to tell what a good cable is ('good' brands included), or if the cable is even the problem.

I like your popup, but half the time it would result in the user finding/buying a fast cable, plugging it in, and getting a “this cable can’t support the power requirements of the device” popup. Maybe we can generalize the popup to just say “nope, try another one”?

Regarding the negative impact of Postel's suggestion, see also The Harmful Consequences of Postel's Maxim:

https://datatracker.ietf.org/doc/html/draft-thomson-postel-w...

In Section 4, they define an alternative design principle based on the lessons of protocol deployment:

Protocol designs and implementations should be maximally strict.

There is also something to be said for too many optional features making a mess out of ecosystem

> Protocol designs and implementations should be maximally strict.

Wasn't this the goal of xhtml?

It is much harder to become more strict after adoption, even if it would bring considerable benefits.

HTML5 is strict too. But it is strict in a pragmatic way, which makes it much nicer to write and read. Instead of forcing authors to write verbose code to an exacting spec without errors, they allowed authors to take shortcuts and make errors. But crucially, then they rigorously specified how those shortcuts and errors must be handled in all cases.

It doesn't matter if all your angle brackets are in the right place or all your attributes are quoted. What matters is that all implementations do the same thing given the same input. You can be strict and achieve interoperability without punitive error handling.

> But crucially, then they rigorously specified how those shortcuts and errors must be handled in all cases.

If the behaviour of such code is rigorously specified, is it even correct to label them "shortcuts" or "errors"?

I feel that this all could be stated as "the price of high compatibility is inconsistent performance."

Which, if I had to rank the two, I'd choose the compatibility easily. Users who actually have a need for 40Gbps transfers will either put the effort into getting the correct cabling, or they will quickly learn that lesson. The fact that they have a working option in a pinch is far more valuable then assuring they can't possibly make sub-optimal cabling choices.

The article has to go to great lengths to even try to make this point. I mean, sure, if you hack up an unlabeled USB cable into a PSU and then plug it into devices it wasn't intended to be used with you might fry the ports. Is this really a problem a "user" is likely to face?

I have a hard time considering any problem that can be solved by adding a label to the product "hard." If every cable was required to have the name of the highest-numbered spec it implements, e.g. "USB-4" or "USB-2.1", both of these problems would either become non-problems or, at least reduced to "go buy the correct cable."

Only if someone enforces that labeled cables are correct. While that is done in theory, I have a USBA to USBA cabel for some DVD drive, why wasn't the manufacture sued to recall all those devices (they don't claim it is USB they just use the USB connectors for their own protocol)

Make it part of the standard. If you license the standard, and you say you're implementing it, that requires you to actually implement it. Done.

You don't have to license USB, you just need a VID. The USB-IF is planning to enforce some of the labels with trademark law, but (for example) for the Nintendo Switch Nintendo often didn't mention that the connector was USB-C, so that wouldn't work.

So? Make USB X+1 require a license to use the name. It ain't rocket surgery.

I tried using a power-only USB cable for data once, then tossed it back in the bin. Later, tried it again, back in the bin. I think on the 5th try, months later, I finally threw it away.

Also, Dial-Up/Ethernet pairs seem to be too complex? Each house I bring my Ethernet/wire Tester+Detector to, somewhere I find a rat's nest of copper cable. Just today I found an Ethernet cable that connects 4/8 to nearby the router, and the other 2 Pairs of wire terminate in a Dial Up cable in another room!?

I threw all of my random USB C cables and Lightning cables away and standardized on these:

https://www.amazon.com/dp/B093YVRHMB?

- USB A and USB C on one end

- USB C, micro USB and Lightning on the other end

- 100W PD

- 20GB data

- video over USB C for my portable second display (https://www.amazon.com/dp/B095GG31KX)

I never have to worry about the “right” cord. I did keep my MagSafe cord for my Mac.

Does that standardisation have 6 cables (including the MagSafe)?

MagSafe is just a nice to have. The MacBook can charge at 100W over the standard USB cable.

Wow nice! I use multi-adapter cables for charging, but always thought the data rates on the enabled connectors would be shit. Happy to hear a review on a product like this that might help; gonna grab a pair.

Cue the mad rush of a hundred thousand counterfeit cables within the week.

Would not be surprised if there was an automated bot on HN and reddit et al that looked for comments like mine with reviews and did exactly that. Damn. Good point.

Thanks for the product tip!

I bought a couple of multi-connector cables similar to the ones you have, but from a different manufacturer and no real specs on the packaging a couple of years ago.

They both charge whatever is connected extremely slowly, no matter what power supply or device. They can barely charge a phone overnight, and charging is so weak that if my phone is used while plugged in, battery net-drains instead of charging. Would've been better to buy cheap, separate charging cables and more power supplies to keep the different device-end connectors powered.

But to be fair, there were no specs when I bought them in a store, just claims that they can be used to charge different devices from USB, which they do, just surprisingly badly.

Fast Ethernet (the 100Mbps kind) doesn’t use those pairs and works just fine with low frequency noise of a landline phone.

Any cable ran in the previous decade should be dedicated to Ethernet, though…

Those cables have one decent use. Charging while out and about with only an unknown USB outlet available. Perhaps something like the USB port on your airplane entertainment system. No data means no virus risk.

USB-C supports data transfer over the CC pins so there's still a potential backdoor.

Add USB hubs to the mix and a true party begins.

Until I need a USB C cable that can carry power and video for my portable external display and almost no random cable that you pick up supports it at all.

So I have to be extra sure that I have the right cable before I leave for a trip.

I'm sure the Quest 2 link cable is theoretically capable of that. However, given that the standard is such a mess, there's a decent probability that the display, computer, or cable won't trust another component and thus fall back to something much slower. It's also kind of hilarious that the Quest 2 needs a separate USB-C cable from the included USB-C charger to actually talk with a computer.

I've had an interesting SSD (SATA) drive failure last year that only my USB2 drive dock could read (fully, no isues)... couldn't access it using anything newer I had.

The OS should detect such issues and advice the user on ways to fix it.

Reply on news.ycombinator.com