Mine is iPhone as hotspot/'net sharing feature, for other mac products (macbook pro, macbook air, etc.), be it bluetooth or wifi, let alone non-mac clients == ongoing failures/stoppages, regardless of iPhone or macbook models.
Back in the day, you flicked your computer's on switch, and it turned on. (And then it took ten minutes or so to load software from the tape drive – much faster from floppy, but still far from instant.) Now, our processors and storage devices are millions of times faster, and computers can take minutes just to start. Software needs to load itself into memory, then initialise the initialisation routines, then actually initialise, then you have to wait for the data to populate, then it's still slow until the cache has warmed up.
It's 2022 and if I want to play a rhythm game, I have to calibrate my input device because, since the days of PS/2, keyboard latency has gone up so much that I can hear the delay between the button click and the computer's reaction.
It's 2022, and when I draw something on one device, it takes several seconds for the drawing to make its way to the other, even if they're in the same room, because the data's being sent through four layers of format conversion in JavaScript, and the traffic's being sent half-way 'round the world.
It's 2022, and "blazing fast" means "slower, and probably more buggy, than Windows 95".
Low latency is still available. Get a fast gaming keyboard/mouse/monitor/NVMe SSD. Use a latency-tuned kernel, e.g. with full preemption enabled. Disable swap and run a userspace OOM killer to avoid the problems this would otherwise cause. Run Xorg with compositing disabled and TearFree disabled. Disable vsync everywhere. Don't use software that caps the framerate low (e.g. terminals using libvte[0]). Turn off all UI animations.
Most software I use feels more responsive than anything from the Windows 95 era.
With all of the extra power we have in our components these days, companies feel like they need to utilize all of it, otherwise it's wasted. Thus the suite of horrible programs required to get a USB controller to have deadzones, or for keyboards to have custom functions.
The amount of background processes in one Windows 10 instance without having opened a single program manually is mind-boggling.
My biggest gripe lately is being forced to move more and more of my daily work off of a very fast and beefy local workstation to an AWS Workspace that feels like I'm time traveling backwards 20 years to a laggy remote desktop connection.
You must have used a different Win95 than I did. Slow and buggy in nearly everything. Had you said Win2000 or XP, I might have agreed more, except that they were really slow. My windows laptop boots up in about 5-10 seconds, and it is rare that something crashes on it.
No, I'm aware how buggy Windows 95 was. I used Windows 98 SE, which was still slow, and would occasionally just get stuck in an uninterruptible "non-fatal" BSOD, hang, BSOD loop before deciding that actually, it is a fatal error after all. From its reputation, I imagine Windows 95 was worse in both respects. I am comparing modern computers to this. Rose-tinted glasses my foot: I am pointing at a computer that would assume the delete key was held (for no identifiable reason) at boot,¹ and I am saying it's more predictable, less buggy, and faster than modern computers.
It took under three minutes to cold boot (~5 seconds from hibernation), took less than a second to load a static HTML page from the 2GB hard drive, and could play tanks.swf[0] for hours without spinning up the fan.
Compare that to modernity: your average Windows 10 machine will normally refuse to cold boot in favour of "fast startup" (basically hibernation) because it takes so long to start normally. If you try that anytime, good luck getting to "I can type with less than a second's lag between a keypress and the character appearing on the screen" in less than ten minutes. If you have an SSD, 16GB of RAM and an octacore processor, you might be able to get that down to 3 minutes – juuuust about edging out the "wait for it to boot, double-click on a document, and wait two seconds for Wordpad to open" of 1999 laptop hardware and software. (And if you give those resources² to Windows 98 it'll boot the whole OS and have your file open in less time than it takes Microsoft Teams to hide its splash screen.)
In a modern web browser, it usually takes less than a second to load a static HTML page. But sometimes it just takes upwards of 10 seconds, sitting there and spinning on a blank page with a blank address bar and (aside from the fan noise) no apparent sign of life.
And… well, good luck with tanks.swf these days; without active cooling, you can feel the temperature difference through the case. My laptop³ went up over 10°C with the fan. A simple Flash game where the majority of the gameplay is stationary graphics idles at half a core, averages one-and-a half cores and peaks at ~2.8 cores (70% CPU utilisation).
I think it's fair to say that computers have got worse.
¹: Yes, it would beep incessantly during the DOS phase of the boot sequence (BIOS beep, then a pause, then a faster, higher-pitched beep I can only assume came from WIN.COM), then proceed to try to delete all the files on the desktop once the desktop loaded. If you timed it right, though, you could tap the delete key before the first beep, and save yourself the hassle of waiting for the keyboard buffer to drain.
²: Minus the RAM, of course; that's enough to overflow its 32-bit address size and make its memory allocator think there's no RAM.
Transferring files between two computers/phones/devices right next to each other, regardless of operating system.
Most solutions rely on sending a file to some server in probably a different country, and then downloading said file on the other computer. Or, using a USB stick to pass the file from one to the other.
KDE Connect works for me. Granted, I've only tried it between an android phone and a Linux laptop, but I've heard from others it works well on other OSes and phones too:
For me transmitting any file over the network between my android(Termux) and my laptop(Fedora) is awful. It starts with a quite fast transmission (Like 60 Mbps), but always drops to just a few kbps after a few hundred megabytes. Netcat does this, python HTTP-Server, FTP, scp, everything fails.
I feel your pain, I tried tons of other solutions before discovering KDE Connect. It works well for my laptop/phone combination, but I'm not surprised if it doesn't work for others, given how horrible each other solution was that I've tried.
edit: I wonder if perhaps some part of your phone hardware overheats after N megabytes because it's not built for constant file transfer, or constant file transfer using these protocols at least? What if you try throttling it on purpose?
I think it is the fault of the network stack of one those two devices, otherwise I can't explain that. Maybe that the devices receive/send data faster than they are able to process. (Is this even possible?)
But I could be awfully wrong
Regarding your edit: That's a good idea, but I never attempted that. I can't test that, because I'm in a Wifi where no device is allowed to talk to another :(
I just use https://file.pizza/ works on any OS with a web browser and the file is transferred p2p so its very fast over local networks. And the Url is easy to remember.
I think croc is a superior solution here. Encrypted transfer. Automatic local peer detection. Human speakable commands. Turns off when you’re done. No firewall fiddling (and unfiddling)
Pythons web server is single threaded I believe so any simultaneously connections break.
I do this but I always get a little nervous about it being accessible to any process anywhere on that network. If the system firewall supports it, I think it would be interesting to make the http connection, and then restrict that port to existing connections only. Once you are confident only your intended client has connected, put the files in the directory.
Or even just a small piece of text, i.e. a link. I have a samba share on one system for files, but that is certainly not "regardless of operating system", as it's a pain to do that with a phone.
So a discord server with just me in it is ... also gives you handy direct links for files up to 8mb.
Taildrop (https://tailscale.com/kb/1106/taildrop/) mostly filled this for me, although this is just for the same Tailnet. AirDrop is also pretty slick. Otherwise I chuck it onto my NAS.
Of course all of these solutions have some serious pre-requisites...
This is the only thing that Bluetooth consistently works for. I can send files between my laptop and my feature phone, and other people's laptops, and other people's phones… It's more reliable than email.
Bluetooth doesn't consistently work for connecting I/O devices to my computer, though. Only peer-to-peer file transfer.
Bluetooth almost got me 100% there as well! Until I discovered that Apple devices don't support sending/receiving files via Bluetooth. But at least it covers about 80% of the devices I own and doesn't rely on external services like most other options people wrote about as a reply.
Ah, sadly the 20% in the amount I wrote about above is one iPhone that won't play nice with the Bluetooth coverage I have at home with other devices :/
Yes! One should be able to take advantage of the speed of a local transfer. One doesn't always have access to a blazingly fast Internet connection and file-transfer service through which to transfer data. One should be able to transfer data between neighboring devices even when no Internet connections are available.
AFAIK, python doesn't ship by default on any operating system anymore (used to on macOS). Otherwise yeah, it's a good contender. You'd still need to pass the URL somehow to the other machine :)
i know you probably know this & you said OS agnostic - but apple really nailed this between macs and iphones. The shared clipboard feels like magic, and airdrop works great too.
Oh I see. I took it to mean there are no platforms where this works, but they make there is no scheme that works across all platforms? The problem with the latter form of the question is there will always be some garbage platform that doesn’t participate.
Dunno, I'm mostly missing the day when we had the technology to create responsive user interfaces. Like imagine we used to have progress bars that actually reported progress, not just some bogus animation.
It's amazing how often things break with absolutely no indication of why or what caused it. Load a page, get a completely blank screen or component. Why? No clue. Everything just silently fails and redirects you to a happy path, e.g. back to a home page or a starting screen. Maybe if you're lucky you'll get an "Oops, try again later or contact support for more help!"
The fashion industry is so rife with this. It seems like any URL that is more than a day old is going to 404 or silently redirect to the front page of the store.
Super frustrating if you read fashion-related reddits. 90+% of links on posts more than a few months old are dead. Only the ones that go to Instagram albums or something like that, are usually still OK. True even for companies that mostly sell "timeless" fashions.
Yep, my workflow is Find something cool in an MFA inspo album -> find it in the comments -> 404. I dont know why I even bother sometimes. It's literally every brand from small botique to the gap, so it must be 'working' somehow, right?
I just miss responsive desktop applications. I see USD10000+ tools running in the browser and I just wish we went back to how solid applications were built in 1990s or 2000s.
I've said it before, but I genuinely think Windows 95 is the pinnacle of UX. It is amazingly intuitive.
The use of depth and colors is streets ahead of the contemporary flat GUIs. You never have to guess the type or state of a widget, or where it begins or ends. Buttons typically have both icons and text. Icons to make it quicker to navigate, text makes it accessible without being fluent in Linear B ideograms.
Why do they need to be smooth? What they offer is an indicator of progress. We can tell that the computer is doing something. Maybe not when it's done, but if the bar inches forward, it's not stuck at least. spinner.gif does not tell you any of those things.
The issue is progress is difficult to report. If I've got Process A displaying my progress bar and Process B doing the actual work I've got to have some sort of bidirectional communication between the two. It's more difficult if B is calling independent processes in the background. They need to report their status/progress to B which needs to broker those updates and pass back up to A.
It's not an intractable problem, just tricky and annoying. If a progress bar is a tiny fraction of an app's lifecycle it's a lot of work to make it accurate for little payoff.
Implementing a progress bar that works for things like "bytes uploaded" or "number of tasks in the queue" on the frontend is fairly trivial. The hard bit is how to transmit a progress value from the backend to the frontend. Typically that requires setting up either a socket or a keepalive connection, or polling an endpoint from the frontend, which is enough of a pain that very few web apps bother.
Ultimately everything required for this sort of feature is right there in every browser. It's the backend that's usually lacking.
Biggest offender is a large OS upgrade on Mac and it sits on that black screen. It often doesn't even say how long is left, but when it does the 'remaining time' is laughably wrong. Not really excusable when Apple have such a small range of hardware configs.
I've seen some Windows installers 'x remaining' consistently count upwards instead of downwards...or the other classic, it sits at 10% for a while and then suddenly jumps to 100
Yep, now you just get a spinning disk, a frowny face and some inane message that "our elves broke something uwuuu" and asking you to contact their (nonexistent) support channels.
I think there's a significant shift from empowering users to controlling users.
Users have become an instrument, rather than ends in themselves. You're attempting corral them so they do something you want instead of empowering them and offering them tools they may need to do what they want to do.
The very notion of power users is deeply incompatible with this.
Yep, while the focus in the past was to appeal to power users (because they would bring in the regular users through word of mouth and recommendations), now that most users are already captive to a variety of platforms, the focus switched to dark patterns and deceptive practices to deepen the lock-in and increase switching costs.
Now power users are merely an inconvenience, since they usually advocate against dark patterns and prefer purely functional software.
I'm not sure how to really combat this. It might be too late.
I should be able to programmatically check my bank/brokerage balances, check when bills are due, and in general automate my finances. My judgment on when to pay what bills is entirely automatable and yet every payday I spend 45 boring minutes doing it myself anyway. Basically, I want to pay all bills that are due before the next payday. If I have money left over, I want to reserve $X in my checking account for spending, put some in my savings account if it is below $Y, and transfer the rest to my brokerage account to invest.
I'm honestly baffled every time I hear about people in the US spending their time literally paying bills. Direct debiting works and has worked for decades in Germany, and also across Europe since SEPA. Same goes for savings and transfer scheduling, no problem at all. Doing "if below X, do this otherwise don't" is not something easily done with bank's own software, true, but I suppose it could be mostly automated by going FinTS.
Direct billing works here too, but I steadfastly refuse to sign up for it, because I would rather decide when it is convenient for me to have the money leave my account. I will never forget that one time back in the '90s when all the utilities decided to withdraw a certain month's payments before the relevant paycheck had actually arrived! Not a good experience. Doing it manually makes it much easier to determine how much money I actually have to spend in my debit account at any moment.
I feel the same way, though some kind of configurable rules to determine when the debit comes out would go a long way to solving this. You can imagine an autopay interface where you say when your paydays are (every other friday, nearest-business-day-to-3-days-after-the-1st-and-the-15th [this is really my spouse's pay schedule], monthly on the 12th, whatever) and then rules like "always take the money the day after the first paycheck in a month, and only if it's less than $500". Bonus points if this gives you extra grace period compared to paying manually.
Fidelity's cash management accounts let you set rules. I have one CMA for general funds where my paycheck is deposited and a second CMA that is configured to keep itself at the value of my next few months of fixed expenses. It will grab money from the general to top itself off after something is paid, and the general in turn automatically sends excess to an investment account if its value gets too high.
Basically the only thing I do is check the autobill payees to make sure none of them are doing anything stupid. I have some trouble with executive function but I have not missed a bill in years because of this system. That might not seem impressive but losing money to late fees is a recurring theme in adhd communities.
Why would you say this? Now I have this idea in my mind of a python script moving my money around for me twice a month and I know damn well that I'll never get to have that.
Open Banking generally mean someone can build 3rd party integrations on top of banking APIs. It is generally not for individuals to get HTTP API access to your own "production" bank account so you can write a Python script to move money around.
I was very happy when it first appeared and my own bank made a big noise about supporting it, so went around to dig for the documentation. Had to sign up for a developer account first, that's all fine and dandy. Write some words about my use case, that's also fine I guess, not optimal but whatever. Finally arrived at the "Success" screen that said something like "Congratulations, we'll review your case and give you access to the sandbox if accepted" which was... Fiiiine.
Some days later, get sandbox + documentation access, write my ~20 line script to move 5 ImagineDollars from account 1 to account 2 within the same holder and want to see it working live but no, here comes a form with ~20 fields describing again the use case, how many users you expect, what you do in case there is errors and so on. I fill it out from the perspective that I'm just writing it for myself.
Wait ~2-3 weeks and receive a rejected email, they don't want individuals to write their own scripts, the service is for products to be built on top of their APIs, nothing more and nothing less.
If you do find a bank that lets you at least query balances and transaction history yourself, I would love to hear about it! Forget about actually initiating any transfers, I just don't want to type numbers into a spreadsheet by hand like a god damn troglodyte.
My small, local credit union lets me download spreadsheet files (not sure if they're CSV or Excel TBH) of all my transactions, I think up to 3 years back.
Presumably this is some kind of off-the-shelf software though it's white-label so I don't know how to figure out what it's called, or how to find a credit union that uses it. That said, I find such credit unions are easy to ask questions of and get satisfying answers: the people actually working there will pick up the phone, rather than farming it out to a call center states or oceans away. They will therefore probably be able tell you if they offer this functionality or not before you open an account.
Change the printer brand. I have a Brother laser printer and once configured the wifi on it, it is zero config. All my computers (linux, mac, windows) detect it over the network. I can also print from my android device.
I've bought an OKI color laser print/scan/copy combo after we last moved some four years ago. Best thing ever in general and I never had any connection problems. But yeah, the mere fact that you have to be choosing your hardware wisely is precisely the problem here.
I'm at the point that I am about to install a dedicated service that people in my house can just forward a document to via email and it will be printed. Then I have one place to troubleshoot instead of multiple PCs and phones.
It's accepted to wait 6 seconds for your TV to start. Washing machine to light up, dishwasher to be responsive, or even apps to take seconds to boot up....
- Why is bluetooth so consistently buggy? Unless you're using the exact bluetooth headphones made for that device, I find myself needing to connect/reconnect or debug so often.
- Why don't TVs still have physical buttons? Sometimes the remote batteries die or I can't find it. Just let me turn things off or adjust volume.
- Why isn't there a sliding toggle "checkbox" in HTML? Why do I have to abuse a checkbox to get one?
I read the v4 spec many moons ago. It’s not that it’s bad, it’s that it tries to define everything rather than stop at a sensible place and let others quibble over the standards.
To your last point; because it should be a touch specific interface but it's abused as a checkbox on desktop by designers. Mobile app platforms have this natively but HTML is supposed to be platform independent.
Humans. It doesn't matter what country or culture. We're mostly busy arguing over the colour of a political party, as opposed to working on the root cause of important issues.
What doesn't work is just sharing stuff between different people and devices. I want to send a web page to my son on his desktop. I want to play a video on the TV. I want to play some music on the main speakers. I want to collaborate on annotating an image with my wife. The whole family wants access to the e-book collection, movies, music, etc.
All these things are possible now to an extent, but each and every one is some degree of difficult to set up, and it's own little thing. And not even mentioning the fractured wasteland that is messaging.
All of this should be baked into the very fabric of computing by now IMHO, a seamless landscape of shared data.
Specifically I volunteer for a group that uses SMS with embedded links for the volunteers to tap to check in and out of a job. Frequently these links are truncated or divided in a way that makes their URL structure invalid. This happens all the time due to the difference in message length and how iOS handles longer messages.
I know this boils down to an SMS character limit kludge, but our Android users do not experience this. The message comes as one block or "bubble" but the iOS users see multiple bubbles that are frequently out of order.
MS Office issues. The whole project back in the 90s as I understand it was that all of these productivity tools would work together. You write something up in Word, move that over into a PowerPoint slide and add in a cool data visual from Excel. I don't know what happened but they do NOT play nice together anymore. Tables don't copy right, I use a totally external program to screenshot the table or pie chart, and maintaining formatting between programs?
Fahgettabouttit.
Word is also remarkably bad with long (100+ page) documents and lots of different content sources. If you're not pasting as pure text into Microsoft Word, you're playing with fire.
Also, I suspect it's cruft inside the program that has made this problem actually worse over time, not just "still a problem". Thank god Markdown is now more widely supported, including on Microsoft-owned Github.
Home routers doing a good job at Quality of Service - at allowing both big downloads and a responsive video call at the same time. This is a common enough usecase these days that it's crazy how a software update on one computer will obliterate a video call on another, even with "QOS" turned on.
If you forgo trying to run a branded home router out of the box. You have to find a brand you can run OpenWRT on, and then install and configure OpenWRT. AKA, you need a somewhat specialized skillset that most folks won't have (it's always worth remembering that only 35% of people are able to complete tasks using only a computer and the internet).
I agree with you, hence my first sentence in the previous post.
In fact, just today a friend made a comment about his Internet that clearly spelled out buffer bloat. I didn't have a good recommendation to make him, and his current router doesn't even seem to have the most basic QoS.
This is not an easy problem to solve though, especially since connections fluctuate which means a static SQM configuration will not always be the ideal.
There are efforts to adjust the threshold dynamically but they're not quite there yet, and are even more specialised.
I've not experienced it but I still find it quite scary: automated account termination with no tangible recourse.
Get your account flagged by a black box and you could loose decades of emails, contacts, bookmarks and favorites,... Then it's a Don Quixote fight to get anything back.
Tried for hours this weekend trying to share files from my Mac Laptop to a Windows 11 PC over the network. Both had latest versions of operating systems.
Could not for the life of me get it working. Had to dig up a usb thumbdrive.
Perceived sluggishness in the desktop response.
Perhaps it's my memory or imagination but I remember that dos/win3.11 had an almost immediate desktop and application response on a pentium.
I'm sure with a lightweight linux I can approach the speedyness and MacOS on an M1/M2 has a good enough desktop response.
It's not just you - we went from "render when ready" to "start an animation of a fixed duration when ready". At least, that's one of the issues I've found.
BeOS was magically snappy and responsive on 7200RPM spinning rust with a single-core 166Mhz Pentium and 128MB of memory. Felt luxuriously powerful. The only widespread current OS that's even close to that level of responsiveness and good-use-of-resources is iOS, and it's been slipping for years such that I now hesitate to make the comparison. And it gets there largely by cheating (suspending backgrounded programs).
USB 2.0 has a theoretical minimum latency of 125us. PS/2 has a theoretical minimum latency of 659us, because the maximum serial clock is 16.7kHz, and the smallest scancode is 11 bits. This ignores considerations of matrix scan rate, debouncing, key travel distance, key activation point, the fact that many PS/2 scancodes are bigger than 11 bits, and the fact that USB can transmit more than one key at a time.
rtings.com measures the keyboard I'm using at 0.5ms latency[0]. I'm confident this is faster than any PS/2 keyboard.
You can even confirm they are all being rendered with `document.querySelectorAll('tr').length` which confirms all rows are in the DOM and it renders smoothly at 60 fps for me.
Database versioning. Why can't I have version control on a database record just like we do for documents? Why do we have to implement audit trails over and over again?
This is the first thing I've seen so far that's genuinely surprising to me given all the machine learning breakthroughs of recent years.
Lots of other things mentioned piss me off too, but I can usually see (to my dismay) where the perverse incentives lie, or the lack of a market drive, or whatever.
This seems like a problem people would like to see solved, where nobody has reason to get in the way of solving it, and that should be solvable by now.
Bluetooth. It sucks in every kind of way possible. It never connects when you want it to. It autoconnects when you DONT want it to. Pairing is a crapshoot and praying.
My best/worst UX: my android phone sometimes temporarily increases the volume to 100% when i connect to headphones via bluetooth, then has the audacity to give me a warning that listening at high volume can be dangerous.
Primarily because it's become infinitely easier to avoid. I don't think I've printed much beyond sheet music in the last decade, and even that was trivially replaceable by a small tablet in situations where I didn't have a Kinko's nearby.
Bought my brother laser printer in 2019. Zero maintenance zero worrying about the INK DRYING OUT (wtf) and my wife can print from the LAN or her phone. I've done zero tech support so that makes it worth every penny. To me it's a solved issue try it!
Bought my Brother laser printer in... 2017? The laser printing aspect is great[0], but I can count on one hand the number of times network printing has worked. I've completely changed network hardware a few times since then, so I'm pretty sure it's the printer.
[0]: Assuming you don't need color or images. Black and white images look awful.
Charging cell phones is more of a pain than charging my cordless home phone.
Even 30+ years ago, it was simple. The phone has a docking station, and you simply place your phone into it.
Why do I have to fiddle with a cable to charge my smartphone? Some phones support wireless charging, but they're rare, and some manufacturers have dropped support for it.
> Even 30+ years ago, it was simple. The phone has a docking station, and you simply place your phone into it.
I don't remember it being so simple. Every mobile phone I had in the 90s/early 2000 had a different charger, even within the same brand. They all had little holes with one pin in them, but all had different sizes, or completely different "couplers". I still have a assorted box with probably 10+ mobile chargers from the same amount of phones.
At least now all Android phones are mostly using the same kind of charger, and Apple being forced to support it as well. Hopefully we'll all be using USB-C for quite some time now.
Not that I have seen them myself anywhere in real-life, but seems there is bunch of charging stations for various mobile phones (at least the ~three different models I've had last years (both Android and iPhones)). Maybe try one of those :)
Control desktop monitor settings still requires using the display's built in controls. This should be directly accessible to the OS so one can quickly lower or raise the screen's brightness like on a laptop.
Manually turn off bluetooth once disconnect from paired device, such as a car or headphones. Save battery and reduced in-store tracking.
Interactions with smart phones still requires multiple attempts to properly format text, fix corrects on unknown words or industry jargon, and copy and paste text.
Android 12+, Google camera no longer follows the standard of allowing to select the photo gallery application and requires open it manually. Older versions allowed for it. Now Google forces you to use their application and break standard. Another Do as I say, not as I do.
Even worse than printing I find scanning. It's like we're stuck with some ancient technology that works maybe 20% of the time. When it works, the scanned page of text is 8MB. Open in Acrobat Reader, export with "reduce file size", 6MB. What the hell?!?
Things that could help:
-Scan as black&white (grayscale) and save as such (8bit instead of color which is 24bit)
-Increase contrast so background is pure(r) white instead of varying grays
-OCR and save as docx/pdf/txt with actual text instead of image of text.
Bluetooth. I’m fairly vested in the Apple ecosystem with a MacBook Pro, iPhone, iPad, and a few AirPods, and never have I seen a feature that creates so much instability across such a wide range of gadgets. I have gotten kernel panics on my MacBook, hard crashes (straight to the Apple logo) on my iPhone and iPad, and had several AirPods refusing to switch devices without a hard reset, all from the Bluetooth dameon. And, don’t get me started on default audio inputs with Zoom calls and whatnot. Rolling dice seems to have better odds than getting the right input to connect from the start.
This is my personal opinion. No offense to any developer or project, but:
The linux touchpad experience... After shifting around responsibilities for a decade between libinput, compositor, desktop environment, applications and back, not many people seem to really see the importance of the touchpad as input device. Especially precisition, interial/kinetic scrolling + rubberbanding and palm detection do not feel optimal.
After digging around with libinput and libinput-config on fedora 36, I have achieved at least an acceptable level of touchpad usability and I'm looking forward to 2023 :-)
The only thing you might want to be mindful of is to place the blame where it lies, not "shifting around responsibilities [..] between libinput, compositor, desktop environment, applications and back". Most of the time the true culprits are the hardware vendors that don't provide technical specifications for their devices, not anyone else.
Open source devs have reverse-engineered the pointer acceleration curves for countless devices and should be thanked that one can use an off the shelf laptop with a working touchpad instead of thinking they're somehow to blame.
I have a term I know I used in some set of documents. I can't remember where I put those documents in some file structure. I use the search function. Friggin' 10 minutes later it finally stops searching, returning a gazillion files which have nothing to do with the term I was searching for.
Yeah, yeah, search is hard on local systems. Turn indexing on. Blah blah blah.
Like, I have endless Halo and ads in the home menu, but I don't have the ability to actually use the computer in any productive way.
I feel the same way about Chrome's browsing history. I was definitely looking at this site 5 minutes ago, why can't I find it in the history?? Plus it only stores like 30 days of history. What's the point of that? I have to install a third party extension to make history usable.
My theory is they deliberately sabotage it so you have to use Google instead to try to re-find what you were looking at.
Windows Search has almost always been useless IMHO. I've used an old copy of Funduc Search and Replace forever for this and when I can't on another machine, it's FINDSTR from the commandline. If you're looking for just filenames, voidtools Everything is awesome.
Gotta add another vote for printers in general. All of our computers "lose" the printer at least once a month, even though it has a static IP address etc. Only solution is to remove and reinstall. Then it never remembers quality settings, which is important because normal quality is utter crap that nobody would ever want. (TBF that last part is more of an OS problem.) Then there's the "can't print black because you're out of yellow" thing, and so many more. How can everything about printers still be so awful even after decades of that awfulness being so well known that it's the subject of memes and jokes?
Special bonus: suspend/resume on even the most vanilla of laptops. The one I'm on now keeps hanging on wakeup because (I'm 90% sure) the fingerprint-reader driver is total trash. I've been hearing lots of reports of suspend/resume problems with 12th gen Intel processors, on multiple platforms. Windows or Linux, same. Again: how is this something that vendors can't even test properly, after all this time?
While not the most "mission critical" application, making basic Siri requests via a HomePod has never improved since the HomePods came out in 2018. Asking to do simple things like turning on a specific light via HomeKit, or asking for the weather, and having the request fail for some strange, unreported issue makes for a frustrating experience.
HomeKit accessories in general have been extremely hit or miss, I had two Logitech Circle View cameras that constantly lost their network connection and had to be power cycled to get back online. I tried many things over the almost 2 years I had them set up in my house and nothing seemed to improve their connectivity. I replaced the indoor one with an Ecobee camera, and the outdoor one with an Eve camera, and haven't had any network issues since. I've seen similar behavior with "cheaper" HomeKit enabled light bulbs too. Certain brands seem to have major issues with connectivity, while others are 100% reliable. Maybe Thread will address this?
The display of sender emails on various email clients and services. It is discouraging to see how blindly clients trust the name provided by a message's sender and display it as-is (and without the actual email address) to the user. The client will naively display
"Amazon Customer Service"
as a message's sender when the full name and email that come in the message are something like
So many technically-sophisticated verification processes are currently run on incoming email to throw out spam and scam messages, yet this particular, rather clumsy form of attack remains unaddressed. Email clients will warn you about everything else but this! I can think of a few simple ways in which clients could detect suspicious sender emails and/or names and help users know that there's something seemingly fishy about a particular sender. Yet I'm surprised that so little attention is paid to this attack vector.
Why the hell is it so hard to just send a file to someone? Why must I go through a central service to host my content before I let someone else view it? Why do I have to tell someone else that they must download third party software in order to just... send files? Without having some corporation store it on their servers.
I assume WebRTC or similar could handle this. I just googled it and this showed up: https://www.sharedrop.io/
But - in many cases people need to transfer files asynchonously - either because they can't guarantee both machines will be online for the duration of the transfer, or the other person wants to receive the file at some other time other than "now".
Email works fine for that case - that's already been solved. It's "transfer the file now" that's annoying. I would love for third party software like Croc to just be included in Windows (like Curl is now) so this isn't made hard. If you can convince someone to download extra software to get your file, you can just use BitTorrent.
There's also https://sendfiles.dev/ for browser to browser transfers. You could also self-host an FTP server (or something more modern like NextCloud) and open a port on your router to let users download (and take it a step further with a dynamic DNS provider in case your IP changes). I agree that it shouldn't be this hard just to transfer a file.
I'd really rather people weren't emailing me large files and if I'm the one sending the file I've got no idea what limits the recipient's mail provider imposes.
No. There are still missing -important- stuff and "works but you have to stand on your hands while singing" kind of things. It's nowhere near being friendly to an end user unless someone picks and sets up the hardware and the software for them.
> and "works but you have to stand on your hands while singing" kind of things.
You mean "works if you don't actively try to break it".
Most stories about Linux breakage involve esoteric distros (especially bleeding edge "rolling" distros or one-man Ubuntu-derived distros) and users trying to "customize" shit they shouldn't.
Can confirm, I try it every couple years and the jank and crashiness/glitchiness (application-level, not the kernel) still make me regret it in a matter of hours. Ran Linux for years on laptops and desktops but just had no idea how much time I was wasting, working around things it couldn't do or fixing broken shit or just living with some stuff not functioning correctly. Every time I go back I find it's the same. Whole foundation of the GUI stack and UI toolkit(s, which is part of the problem) need a tasteful, high-quality ground-up rethink and Wayland sure as hell ain't it.
Linux is problematic on hardware designed for other operating systems. My Librem 15 works flawlessly, including suspend and WiFi (without even any binary blobs).
The biggest thing keeping me from using Linux for my desktop pc right now is lack of HDR support.
Apart from that the whole display servers world is just not working well. Some things still don't support Wayland, you can't share screen easily, X has it's own issues etc. HiDPI is still problematic with both; you get weird scaling issues with apps and and the occasional humongous mouse cursor. Using multiple displays with different DPIs you again get scaling issues. Setting up multiple displays still requires fiddling with xrandr, xinerama etc.
Just let me know if you need more examples. I tried to fully switch to Linux many times spending lots of time + money (bought one of the "recommended" devices).
> The biggest thing keeping me from using Linux for my desktop pc right now is lack of HDR support.
HDR is just a marketing term for software that increases the colour and black saturation of content beyond the actual input. For photography that means more saturation than what you see in real life, for video that means tweaking it after you receive the signal.
An OS can't "support" or not support HDR; if a video has HDR and your display has the dynamic range then you'll see it in "HDR". Unless you actually want your OS to post-process everything on the screen to have marginally higher contrast than normal?
> HDR is just a marketing term for software that increases the colour and black saturation of content beyond the actual input. For photography that means more saturation than what you see in real life, for video that means tweaking it after you receive the signal.
Not even remotely correct. It is definitely not a marketing term. HDR is set of specifications and formats for capturing, storing and displaying visual data with a higher than "Standard" dynamic range. It basically carries more data and it requires proper color management. It has to be supported by relevant drivers and the display server. Right now there is no way to view HDR content (video, photos or games) under Linux.
If you are interested; there is some effort going on to enable this but it will probably take a couple more years.
"HDR" is just metadata to pass along to the device to tell it how to optimally adjust the colours (or not adjust them). It's main use seems to be so that HDR displays can oversaturate "SDR" content in HDR mode.
> carries more data
Not really. The metadata isn't more colours or something, it just passes info to the device about certain aspects of the video.
HDR content will be HDR if the display has enough range regardless of whether or not the OS has HDR "mode". It's more about adjusting all the other content.
Dunno if you ever had a plasma TV back in the day but they could display way more contrast than LCDs of the day so you could adjust the settings to change the way all content looks. HDR metadata is basically that, but on a per-video basis. Useful enough but not some game changer... Which is why there's tons of articles about what HDR does and doesn't do.
Edit - where HDR really is great is photography... You take multiple photos of the same scene with different ISO and shutter settings then stitch them together for more colours than would otherwise be captured. But a screen can either display colours or it can't.
You're getting somewhere but still not complete. HDR formats we have today use 10bit or 12bit bit depths so that alone is a big difference. About the metadata, HDR10 uses static metadata meaning whole content will have the same metadata but HDR10+ and Dolby use dynamic metadata.
HDR content "will not" be HDR if you're playing it under Linux. It will however be HDR if you play it under Macos or Windows (keeping everything else like the screen and content same). There is no way around this right now. So in basic terms you will not see the dark blacks and bright whites under Linux.
> Edit - where HDR really is great is photography... You take multiple photos of the same scene with different ISO and shutter settings then stitch them together for more colours than would otherwise be captured. But a screen can either display colours or it can't.
It's the same for video or games really. With HDR you can record and display more details and contrast.
And that HDR photo you just described will not be shown as HDR even if your display supports HDR if viewed under Linux
HDR as we know today always meant 10 bit colour. Before that some screens emulated HDR in 8 bit color by doing dithering to -poorly- create the illusion of a wider luminance range.
edit: and 10 bit color support on its own is not enough for HDR. I mean I don't know why we are even discussing this honestly. HDR under Linux is not supported at all. It's not an opinion; it's an objective fact which you can confirm in a couple of different ways. You can not view HDR content if you're using Linux no matter what content or display you have. There is an active effort in the community to have HDR support in Linux but it will take a couple more years probably.
Multiple displays have worked out of the box for me for over a decade. Nothing fancy here, just X11 with the plain jane MATE desktop (and Gnome 2 before that.)
Multiple displays works well if both displays have the same or similar dpis. When they are different like one is hidpi and the other one is not, you get scaling issues.
I second the "no". I have given Linux 3 tries (several months, and a lot of patience) over the last 10 years and each time I've had to ship back to windows at some point. It's not there.
Ever checked in on the youth lately? It's all smartphones, game consoles, iPads.
Students will use MacBooks/Laptops/Chromebooks for school if their course requires one but "devices" are the main computing device nowadays. Heck, even many non-technical business people have switched to iPads. Kiosks? Tablets.
The PC is no longer "the" personal computing device...
1. I still can't seamlessly stream from Spotify/android phone to Google/Nest Home speakers. Takes up to 30 seconds to sync, randomly breaks down. I switched to iPhone two weeks ago. Spotify/iOS connects to Google Home immediately and works well.
2. I keep giving different variants of the same command to my Google Home, which often gets answered with "I have no idea how to do that", or sometimes minute grammatical details change the command for the computer, or sometimes it refuses to remember a task that it did the day before with that exact phrase, or sometimes it randomly does something completely different. Why does "Turn living room off" do something different from "Turn off living room"?
I am slowly moving away from Google's ecosystem because nothing ever works seamlessly.
The Spotify Android app is at a minimum in the top 3 lowest quality pieces of software I've used in my life (been using computers since 1992 or so). Probably #1, I'm just hedging in case I'm forgetting something.
Where to even start with this... Ok, how about with Microsoft Outlook? The "undo" after deleting or moving an email works about 10% of the time. Sometimes it recovers the message, but most of the time it either does nothing or even better, brings back a totally different message!! smh
Selecting and pasting text on Android. It works completely differently in different apps, or different text boxes in the same app, despite being a service provided by the OS and not the app. And even at it's best it's a nightmare.
Sometimes double tap will select a word. Sometimes double tap will show the clipboard menu. Sometimes double tap will show the clipboard menu, then quickly dismiss it again. Sometimes to show the menu I have to single tap, but sometimes I have to long press.
If it's a single-line text box, I can't tap anywhere within it to get the clipboard menu, I have to tap (or hold, whatever it is today) on the tiny part at the beginning where the cursor is.
Then there's the whole nightmare of actually selecting the text you want correctly.
Pasted the wrong thing? There's no undo. Just suffer.
1. Devs incorporate an OAuth code based flow. Many apps do this. Passwords should rarely need to exist on a TV device.
2. Paste using the remote app if you have an iPhone.
The first is caused by dev laziness (and possibly better for rare scenarios where there isn't a second device) and the second is perhaps a cost thing (do you have an iphone?)
The typing experience on AppleTV is actively user hostile - the password is masked (you cannot reveal it) so if you even think you've made a mistake in one of your 25 characters, you need to start over.
The second doesn't work for me: we only have Android devices. It's unclear why our 3 MacBooks aren't sufficient for this purpose; I didn't have any suspicion that this would be the case.
There's so much software written these days that is duplicated. We focus on the large framework style pieces (and in JavaScript land rewrite them every year ;)), but ignore the long-tail pieces that are common but domain specific.
This would mean standardized software development building blocks for domain models. E.g. a standard model of money/bank account/user/address/company/shopping cart/food product/ingredients/calendar/... for every programming language.
I've seen a a database-centric version of this (I don't recall the name of the site though), but not a domain model version.
Speech to text is still really bad, Cloud costs are real bad too, checkout https://text-generation.io or OpenAI whisper which will fix that fairly soon I think. Over 8x cheaper than Google.
There's lots of other cloud cost stories like CloudFlare R2 egress saving people millions on large file storage.
I'm surprised decentralization hasn't hit cloud providers as hard yet either.
With that hotspot example my Telco in NZ (spark) blocks me from using a Wi-Fi hotspot on one of my android mobiles, which is ridiculous that they can turn off an Android feature on my phone for profit.
Two brand new Bosch ovens side by side, both showing different times on their digital clocks. Why don't they auto-adjust with daylight saving or at least synchronise with each other. I'd rather have no clock than two wrong ones.
Automatic faucet/dryers/soap dispensers in public toilets. Always delays before they start. Or don’t start at all. Or start/stop before you’re done. They seem to work as (un)reliably as they did 20 years ago.
I still have to fight with it all the time, and cannot get it to work as expected.
And that's with the most standard use cases:
- System always getting muted after unlock/reboot (for some reason)
- Playing sounds on the speaker despite jack headphones being connected (I have to manually switch the output)
- If I unmute the sound (that wrongly got automatically muted) with my earphones connected, it starts loudly playing my music on speakers at maximum volume.
- UI not reflecting the actual situation (it shows muted speakers despite playing stuff on it for example)
Shit, I'd be happy if the Mobil stations near me would correctly implement chip cards on their gas pumps. Every friggin' time, they ask for a PIN and refuse to take "no" for an answer. Every. Time.
It's a credit card. It doesn't have a damn PIN in the US, and I have to go inside to pay for gas every time. I've quit buying gas at the area Mobil stations because of the hassle. Most of the other chains seem to get it right.
How is it that Europe has been doing chip cards (with PIN) for decades, and we can't get this right?
I inquired with the issuer. I can set a PIN for a cash advance, but not a regular sale. Which is fair, I guess. They are fundamentally different transactions with different interest rules, etc. Nonetheless, it's a very frustrating situation that the solution almost exists, but can't be used.
Companies still ignoring business. I tried to get a company account at getsby, nothing happens. I email them, silence. I'm pretty sure I'm not blacklisted by MC/Visa because I have both from my bank and my company has way more money than it knows how to spend.
Oh, and Paypal's payments not working 50% of the time. Probably something with using firefox or being on linux, or having an adblocker. I have a secure password. I use 2fa. It still fails so often, I now avoid it.
Connecting external screens or projectors to a PC. It never made sense to me that the devices on either end take many seconds to realize that something was plugged in.
"Instant edit" on most website. Similar to a wiki but it should go to the user/admin that made the content and he can publish or discard it.
Ofc it would have to be without login/captcha etc.
Every blog post would have every reader be able to correct small mistakes, fix broken links etc. etc. Basically a distributed maintenance system for the web where stuff that is actually used by people is keep up-to-date.
We have trouble keeping Jira and Bitbucket synced. Both cloud products. How can they struggle with integrating themselves?! (Our GitHub repos sync just fine.)
It's no easier to send a confidential message using the most
widespread communications protocol than in 1980.
Searching for anything serious/interesting I pause and think: Do I
have it locally? Else, which specialist databases or sources I might
try. Else use a bash script that hits a bunch of "alternative" search
engines. Finally, use Google/Bing as last resorts. What happened to
search?
Recently installed Linux Mint, what baffled me in 2022 is that regularely upon log-on it nags me to install security updates and when I acknowledge the request, it asks me for my password again. I agree that general package manager operations should be a privileged operation... but approving the installation of security updates should not have the user being greeted by just another password prompt.
They're exactly the same filesystem operations, it's just the semantics of why they're being installed that changes.
What frustrates me about updates is that they nag at log-in when I've already been waiting for the computer to boot and just want to get on with the task at hand.
I recently installed Debian Unstable on a new PC and it came with Pulseaudio installed by default. Just like it did when I first tried Pulseaudio on a different computer many years ago, the audio intermittently desynced from video playback on Firefox. Thankfully, Debian compiles Firefox with the optional ALSA support enabled, so I uninstalled Pulseaudio and everything worked perfectly again.
Multipath mesh networking in ubiquitous devices. We all walk around with a super computer more powerful than the original devices hooked up to ARPANET. Assuming that we believe that E2E encryption is secure, making these talk to each other shouldn't be forefront rather rather than them having to communicate only to centralized hardware (cellular base stations etc.)
Buy once, use everywhere media - we regressed from owning the vhs/dvd/bluray to finding which service will allow you to rent / license the media.
While I'm there, the path from media rental to "ownership". If I watch a movie for $4.99 and it's good, why not take the $4.99 off the price to own it rather than making me pay full price?
Why isn't there a single model for sharing video and alerts to a local PVR? Why do so many crappy cameras require internet access and 1st party apps?
I want cameras, I want alarms, I want heating and lighting automation, and I don't want to share access with some kid in China, or have nine different apps.
Buying a dozen eggs and not having one of them be cracked when you use them. The clear plastic cartons might help you see a fully smashed egg, but it doesn't let you see the small cracks. Especially for brown eggs. The cashier double checks the eggs anyway, but there's always a cracked one once I have paid.
My MacBook Pro 128 GB has "Pro" in its name but it takes a whole night to upgrade Xcode to a newer version on this model if I just pick Upgrade in the App Store (I no longer do it and download it manually instead, and then delete the old version manually before installing the new one).
Sharing WiFi connection details via a QR code in iOS. On Android it's intuitive; but last I checked a few months ago on my spouse's iPhone, it was a weirdly complex process that didn't leverage the information the phone already had to connect to WiFi.
Digital signatures for the wider public are still very cumbersome. The number of times I have had to print a filled PDF form just to sign it manually, scan it, then upload it back to the issuer's website is too damn high.
AirDrop – how is a 50% success rate remotely acceptable?
And my personal favourite – macOS carries over the input language from an app to another, which makes no sense at all and is frustrating me for 17 years now
A goole calendar view with 2,3,4 month or even custom view showing events in the day. Its not like there isnt enough screen real estate to fit it all in with modern displays.
Cross-platform user interface libraries, including desktop<->mobile. GTK and QT work for desktop-only apps, but once you include mobile there's no good, solid solution for this except a web view.
Clicking 2 devices together does not pair them, or at least initiate pairing. Have to scroll through a list of 100 cryptically named devices without any indication which I should connect to.
Trying to find a setting in Windows' 3 or 4 generations of control panels. Honestly, it's amazing to me that Windows is such an unpolished turd after 30 years of development.
Printing something from my computer. I can connect to servers all around the world and for whatever reason, I cannot predictably and reliably connect to printers in my home or office.
Phone / TV / Computer integration that is not just handoff, but treats the 3 things as part of a whole UX. Why am I streaming from a website to my phone to the TV?
Nah, just use a 3rd party controller. You lose functionality on a console (for example, NFC reading and HD rumble on switch 3rd party controllers), and they can't even begin to accept that you might have different button labels than their first party controllers.
Equivocating a helicopter a flying car is akin to calling a Model T a Tesla with FSD.
And those "millions" of reasons all boil down to R&D costs. Reliability, inspections, takeoff-to-landing autopilots (they exist - have existed for over a decade - for military helicopter drones), etc. are all cost issues which are slowly being worked on.
The progress could be faster if it didn't rely on angel and startup-style investments and the associated NDAs and trade secrets, but I don't expect that to happen in today's political environment.
Problems with those I-can't-believe-they're-not-illegal littering rental services, not the bikes themselves, right? Surely people aren't buying ebikes and leaving them all over the sidewalks.
Most drivers (in the US, at least) have trouble enough navigating two dimensions. Adding a third would undoubtedly make matters worse. Hell, just look at the safety record of private jets.
we have only 3 browser vendors left, and they mostly cooperate with each other, and still the default styles are so all over the place that even simple pages often end up with broken features between browsers on the same platform
Closing the lid on a Windows laptop and being 100% confident it won't decide to wake up either in the middle of the night thus feeling the room with an eerie glow - or in your backpack thus cooking the motherboard.
I'm sorry to say this is an issue everywhere. I've had a mac BSOD from unplugging a dock (doing it while the computer's suspended is/was a surefire way to trigger this particular bug), and linux is even more picky since it depends entirely on the hardware underpinning the dock.
To be fair, pluralisation is hard. If you want it to work in English, but you want your software to support translations, you have to call a function and give it all the information necessary for all languages you support. That includes:
• are there 3 or 4? (Plural isn't just zero/one/many in every language.)
• is the quantity small-ish, or large-ish, compared to normal?
• what's the grammatical position of the noun you want to pluralise?
You'd have to have something a lot more sophisticated than string substitution. And there isn't really a pressing need for it. Day(s) works fine.
I think you have the arrow of time backwards. Technology seems to be getting less reliable, not more so. Every time my laptop manages to connect to a hotspot it's a small miracle.
iPhone hotspot over wifi has always worked flawlessly for me, with all types of clients from macbooks to imacs to chromebooks and chromecasts and even my car.
That feature is nice, when it works, but like its ctrl+v counterpart, 20% of the time it just doesn't. Same problem with with ctrl-shift+p, which is supposed to paste just the plain text - a lot of the time, seemingly at random, it just gets into a state where it pastes the formatted version regardless, and I'm forced to paste into notepad and recopy to lose the formatting.
Back in the day, you flicked your computer's on switch, and it turned on. (And then it took ten minutes or so to load software from the tape drive – much faster from floppy, but still far from instant.) Now, our processors and storage devices are millions of times faster, and computers can take minutes just to start. Software needs to load itself into memory, then initialise the initialisation routines, then actually initialise, then you have to wait for the data to populate, then it's still slow until the cache has warmed up.
It's 2022 and if I want to play a rhythm game, I have to calibrate my input device because, since the days of PS/2, keyboard latency has gone up so much that I can hear the delay between the button click and the computer's reaction.
It's 2022, and when I draw something on one device, it takes several seconds for the drawing to make its way to the other, even if they're in the same room, because the data's being sent through four layers of format conversion in JavaScript, and the traffic's being sent half-way 'round the world.
It's 2022, and "blazing fast" means "slower, and probably more buggy, than Windows 95".