Bluetooth. Like it works most of the time, but especially randomly the microphone does not work

It depends on the complexity of the operation. “I want to rename all my files to have underscores to spaces”, CLI will let you construct that easily. I want to move all mp4/mkv files to one folder, but all ‘.opus/.mp3’ files to another folder, CLI is a bit quicker. Or I want to take the audio tracks out of all these mp4/mkv and then name the result according to the basename of the original file and move the result, well, mkvextract and mv are quicker than trying to wrangle all the content in comparable GUIs.

But yes, if you are wanting to do an operation on a file or a range of files easily handled with shift-click to select, then GUI will be both approachable and quick.

I can appreciate the desire for “you know what I meant” CLI interaction, but shudder at the verbosity of natural language in a lot of these cases.

Retaining that much detail on tentacles takes some drive space

Linear density could also boost throughout. Multiple actuators also exist.

Well, Microsoft didn’t offer it to you freely…

Yeah, but without learning Microsoft, how would you know that ‘dir’ just makes sense? Or that you might want to look at ‘diskpart’ to look at your drives?

I mean, diskpart and dir don’t make especially any more sense than lsblk/parted and ls. A fair point can be made for ‘copy’ being more intuitive, but ‘diskpart’ means you had to learn what disks and partitioning were, and lsblk means you need to learn what ‘block’ devices rae, and of course ‘parted’ references partitions. ‘dir’ means you wanted to ‘show the directory’ which means you had to learn of it as a directory, but then learn that the shortname of directory is the way to see the contents of a directory. ls means you learned you want to ‘list’ contents and that unix had this laziness of just the first and third letters of a word. Both involve learning, neither is ‘intuitive’.

You end up writing ridiculously long commands

I assume this is the likes of dbus-send and crap, and I agree with you if that’s the case. Dbus is a complication I could do without and have to confess that powershell cmdlets generally do a better job of instrumenting the system than a system that increasingly has no specific help and only long dbus-send commands to tackle certain things. dconf has issues too, but I think does a better job than the Windows registry at analagous function.

Well it’s not a problem in arm environments generally, just in x86 land, and the bootloaders don’t have a good shot of figuring it out on behalf of the kernel either… There is an ACPI table but no one cares about it in Linux land and is almost never used in Windows land (EMS does support it) and as a result most systems don’t bother doing it at all.

I have trouble getting real displays working all the time…

Needing to know which serial port is which and manually tellling the kernel via console=ttyS0,115200 or whatehaveyou is annoying…

What other displays could people mean?

Same for me. In Linux, I plug in USB-C and both monitors in the chain light up every time without thinking.

For some reason, dual boot into Windows and it always disables one of the two by default until I manually go in and tweak it alive, and then it will do it again next time I plug in.

Now back in the day, futzing with XFree86 config files and CRT monitors and absolutely lots of ‘voodoo’ to match what Windows pretty simply did with display configuration. But nowadays at least with kwin wayland compositor on nVidia proprietary drivers, it always does exactly what I expect without asking, and Windows is the one that assumes that I don’t want to use all the displays that are connected.

Windows seems pretty clunky by comparison nowadays when it comes to display configuration.

Now juggling my bluetooth audio… I think Windows still has the advantage. I have no idea why sometimes my bluetooth microphone just doesn’t work under Linux. I do appreciate the ability to manually select the bluetooth codec in Linux where in Windows it ‘guesses’ and often guesses wrong, throwing it into ancient headset codec territory when I’m trying to listen to music, because who knows what has made Windows think the microphone device is open…

Networking… Linux wins hands down with VPN connectivity, much much easier to manage all my VPNs in one place in the ‘casual’ user scenario instead of a litany of competing ‘endpoint managers’ in Windows. When VPNs step on each others routing tables, well no OS makes that easy but at least Linux network namespaces makes it possible for me to have multiple network ‘worlds’ in one place to reconcile the conflicts…

Probably the other area where Windows has a bit of an advantage is a consistent binary driver model. In Linux if you are an out-of-tree driver, it’s going to suck to keep up with changing in-kernel APIs to keep your source compatible, let alone have a module running without a recompile after a minor kernel update. I guess the silver lining is almost everyone decided to have their drivers ‘in-tree’ to make sure they are maintained and don’t need a lot of ugly #ifdefs to contend with multiple kernel behaviors… Then there’s nVidia and some commercial filesystems that either cannot or will not go in-tree…

Getting a dns name is straightforward enough, and let’s encrypt to get a tla cert…

But for purely internal services that you didn’t otherwise want to publish extremely, the complexity goes way up (either maintain a bunch of domain names externally to renew certificates and use a private DNS to point them to the real place locally, or make your own CA and make all the client devices enroll it. Of course I’m less concerned about passkeys internally.

You don’t have to target every distribution, target a vaguely credible glibc, and of course the kernel, and you are covered.

As a distribution platform themself, they don’t have to sweat packaging N different ways, they package the way they want. Bundle all the libraries (which is not different then the way they do it in Windows, the bundle so many libraries).

They don’t get the advantage of the platform libraries and packaging, but that is how they treat Windows already because the library situation in Windows is actually really messy, despite being ostensibly a more monolithic ecosystem.

Broadly speaking, the private keys can be protected.

For ssh, ssh-agent can retain the viable form for convenience while leaving the ssh key passphrase encrypted on disk. Beyond that your entire filesystem should be further encrypted for further offline protection.

Passkeys as used in webauthn are generally very specifically protected in accordance with the browser restrictions. For example, secured in a tpm protected storage, and authenticated by pin or biometric.

For ssh, ssh keys.

For https, webauthn is the way to do it, though services are relatively rare, particularly for self hosting, partly because browsers are very picky about using a domain name with valid cert, so browsers won’t allow them by ip or if you click through a self signed cert

Because using a gui is easier for novices

To the extent that is true, then the novice doesn’t end up asking for help. The goal is that the capability is discoverable. Or if it’s really a bit harder, there are hopefully tutorial youtube videos that cover the use case.

But when a user asks for specific help on a task they couldn’t figure out for themselves, or are asking for help with a ‘something went wrong’ dialog box, well helping with CLI is much more feasible than the involved mess of trying to basically make video tutorials ad-hoc, or screen share, or try to help them find the obscure log file an application wrote somewhere with the real error.

It’s relevant. From about 1995 to 2006, Microsoft was pretty much hard set on ‘cli is dumb, do nothing cli wise, cmd is a concession, but a crappy one’. As an artifact of that, you got regedit, a godawful ‘GUI’ that took a messy datastore model and just kept it ugly, in a way that would have been pretty much better as a CLI interface.

Microsoft started getting the idea again, but in true Microsoft fashion, had to reinvent the wheel and did PowerShell to try to create a CLI ecosystem from scratch rather than trying to build anything vaguely familiar. To their credit, for first party stuff they did a fine job enabling it, though third party applications remain a mess to this day. It does highlight that even Microsoft figured out that CLI actually does make sense a lot of the time.

Hey, and PowerShell commands aren’t GUI, bringing us back to the point of this post…

If it’s “oh, you can open up [application X] and it’s easy to figure it out, and there’s videos out there to cover your use case”, then ok.

But if it’s to help a user with a very specific task and they want their hand held, well from a GUI perspective I’m either making a bunch of screenshots or maybe even a tutorial video or a screen share session… Or I shoot them a relatively short CLI command that does it and move on to other things.

It is usually much shorter to tell someone the CLI to do something than it is to try to train them on a GUI for the same thing. If it’s well-trodden subject matter, well they probably already found a youtube tutorial and didn’t even have to ask.

Exception for helping someone who sshed into something and doesn’t understand what they are doing.

It happens that someone without knowledge has no idea how to interactively edit a file on a system they can only ssh into. ‘run nano’ is easier than ‘ok, now I’ll show you how to WinSCP the file down edit it, and put it back, but make sure you don’t screw up the CRLF or permissions in the process…’