Originally posted by ldo17
View Post
CLI and GUIs have both pros and cons that make them suited to different use-cases, just saying "the CLI is the one true timeless computer interface" (while ignoring a bunch of CLI pityfalls) only shows the writer is a goddamn moron.
The main reason GUIs exist is because they make easier to do stuff you do rarely, therefore make the user more self-sufficient. Yes you have the offline manuals in most distros but for many tasks you still need to read for quite a bit of time, and many caveats are only in online tutorials. If to use a PC I need to look at online tutorials, might as well have a Chromebook (that is completely useless without internet connection).
For example, I rarely need to repartition drives and I don't feel like reading up how to control that monster called Parted by command line every time, that's why I have Gparted that offers me a simple GUI to control that.
Same for Grsync for movig data around. It's basically a GUI that then assembles a command line argument and passes it to rsync, but it's far faster to remember or operate than remembering all options myself or reading up the docs every time I need it (and no it's not a repetitive task, I'd use scripts for that, copy-pasting the command I see Grsync assembles with my options).
GUI also offers safety rails, for example I could write disk images with dd, but I usually don't and use OpenSUSE's disk image writing tool instead. Why? Because if is fuck up with dd I can nuke the system or whatever, while most disk image writing GUIs auto-hide system partitions, and clearly name all drives they detected as eligible for disk image writing.
That guy shows he has no fucking clue when states bold bullshit like "users’ needs grow with time", and "Android devices which have a UI completely different from the PCs that they were familiar with".
All modern GUIs are still basically evolutions of the good old Desktop concept invented a long time ago by Xerox (and made more popular by Apple, then flat-out cloned by MS in the first Windows).
Icons are still icons, buttons are still buttons (with text or icons on them), text fields are still text fields, Start menu = App drawer, you still have desktop icons or the desktop is the app drawer (like Apple does), top bar (maximize/minimize/close buttons) = bottom bar or hardware buttons in mobile, and so on and so forth. Material design or Windows 10 does not change that you still have the same basic elements in the UI that you had back then on that shitty GUI-for-DOS that were first Windows releases.
At most they move buttons around.
I also find especially funny how he pulls MS's Win8 UI fiasco as an example, there the issue is that some idiot placed a UI designed for a touch interface on a desktop that has keyboard and mouse, not that "overall GUI design grows in complexity" or "problems with users who were accustomed to the old way of doing things". People have 0 issues with using phones or tablets with Windows 8 (or win10 touch mode, or win10 phone).
Sure you have many people that can't fucking use any devices at all and only learn 2-3 tasks without understanding how to read and operate the interface, but that's their own limitation, even chimps can learn how to operate basic GUI systems.
The main reason CLI still exists is because it is great for maintenance (you can see more info in text form than in a GUI, this is also easier to send over to a tech somewhere else like by mail or in a forum), can be easily automated, and uses little resources.
It's great and all, I use CLI all the time to control embedded shit and servers, and wrote many thousands lines of scripts for various dumb repetitive tasks.
But he is very exaggerating CLI's usefullness, most end users need a tool that helps them navigate the interface and keeps them from shooting themselves in the foot, not to automate repetitive tasks, nor to do mainteneance on stuff. There is a reason why Windows's CLI sucks ass hard if compared to even crappy Linux/Unix shells (and PowerShell is a weird beast).
I'd also like to dispell some of the bullshit he says about CLI:
-there are quite a few different shells, Bash isn't the only one and not all have the same syntax and commands (especially Linux vs BSDs vs macOS), yes they are all evolutions of older stuff, but it's the same situation most GUI have, basic concepts are the same but you need time to adjust to use zsh if you always used bash, ash in busybox may or may not run as Bash enough, dash in Ubuntu/Debian init scripts has (had? are they still using it?) its own quirks too
-CLI is clunky as fuck. Yes, if you need to do any truly complex work you need to upgrade to a true programming language like Perl or Python, that is still scripting, but can leverage far easier ways to do shit that piping hundreds of outputs from CLI tools around to get a result.
-CLI is also legacy as fuck. Yes, most of the actual system infrastructure that runs CLI is ancient, barely maintained and poorly documented. So far none had the real need to go and modernize it a bit. Also the whole framebuffer issue with linux kernel shows this, none really seems to give a fuck about making a CLI system that runs with modern interfaces.
-CLI's way of doing things is fragile by design, as it was not intended to be used to do very complex tasks where security and stability are paramount, there is plenty of undefined behaviours, quirks and so on that make it unreliable for true work, this is one of the reasons behind systemd, where our Lord and Saviour Pottering gifted us a far better tool that suffers from none of the drawbacks of scripts when doing critical tasks like daemon management.
-CLI is not low(er) level than GUIs. Not. One. Bit. It is text-based and is different from GUIs, but it's using most of the same pie-in-the-sky levels of abstraction used in GUIs too. So no, you aren't more nerdy if you use CLI, you aren't "closer to the machine" even if it looks more like matrix.
Leave a comment: