DOS vs. Linux

[bostonman]

I learned DOS way back before Windows and picked up on it quickly. Today I find it's still easy to navigate and remember commands. I've also used Linux (or Unix?), however, I find it's extremely confusing.

Linux seems like so many hidden commands. Recently I used DD to image a drive, and it took a very long time to figure out if my drive was connected, which one it was, and kept having command errors.

Also, I find things like updating and/or installing packages to be confusing. Anytime I want to do something, it seems I need to install a package in which case I find the steps online. Once I execute the commands, I never understand what I've done, but know it works.

My feeling and understanding is that Linux is much more powerful than DOS, but I find myself confused over what commands exist, what they do, and of course, things like apt-update....xxxx

Where do the apt-updates come from, and how do I know if I'm installing a virus?

[bob91343]

Perhaps it's time to take a short course to straighten out some likely misconceptions.  I am not an expert but when you run into this sort of difficulty it's probably because your understanding is too limited.

[james_s]

Linux is FAR more powerful than DOS. This should be obvious when you consider that Linux is a clone of Unix which was what ran on large mainframes in the era when DOS was released for the far less powerful desktop PC. Since Windows really took over when DirectX for Windows 95 started to allow gaming on Windows the development of DOS pretty much froze and it does pretty much what it did in the 80s. In the meantime PCs and other devices have become so powerful that running a full Unix-like multiuser OS is no problem.

I cut my teeth on DOS and used it for many years, well into the Windows era but over the last 5 years or so I've done more and more on Linux and now when I use the command prompt on my Win7 laptop I really miss all the stuff I can do from a Linux terminal.

Apt updates come from repositories that are stored in a text file /etc/apt/sources.list, the default repositories are vetted and safe, although you should exercise caution when adding random repositories you find. That said, viruses in open source software are very rare because being open source it would not take long for someone to find it.

I second the recommendation to read a book or take a course, depending on what works well for you. There are a number of Linux courses on Udemy, and almost all of their courses go on sale frequently for $10-$15 so never pay the inflated full price. Lots of tutorials online too, and random youtube videos, do whatever works for you.

[ataradov]

APT specifically installs things from a repository. There is a limited list of people that maintain packages and it is not that easy to sneak in a totally malicious program.

But some programs are legitimately destructive by their nature. Formatting a drive is what drive formatting utility does. This is destructive, but you know ahead of time what you are installing.

It would be virtually impossible to put something virus-like into the repository.

Things get different when you download some random *.dpkg package, as a lot of commercial software is distributed. Then all bets are off, it can be anything and you have to trust the publisher. This is no different when you download a random *.exe installer on windows.

And it is better to understand what the commands do. Some advice you will find on sites like StackOverflow is just bad. Generally you want to find out what package contains the program you need and install it yourself using apt-get. 

[Ed.Kloonk]

https://www.linuxquestions.org/questions/


 

[Berni]

Yep linux is far more powerful but has a different target audience compared to DOS.

DOS (and later on Windows) is designed for the average PC user. Sure back in the DOS days you still needed to know a fair bit about computers to actually install and configure DOS, but it is something that someone determined enough can get trough with the help of a simple manual. It is made to "just work" and the main task of the DOS back then was to run software and that is it. What it could mostly do is file management since that was required to actually get the executable to where they could be run while also providing basic IO to them. And this is what most of PC users wanted. They just need DOS as a tool to be able to run WordStar or there favorite DOS game.

Back then it was expected from the PC user to have to know about computers in order to use them, and DOS went for setting that bar fairly low so that as many people as possible could use it.

Linux on the other hand is a OS made by geeks for geeks. Offering some very powerful functionality at the cost of also being so much more complicated. For someone that is mostly going to just use the OS to manage files and run programs all that complexity is just a hindrance rather than a help. None the less these days Linux is everywhere, so it can be worth it to give it a shot and learn some of it, its much easier to use now.

[Mechatrommer]

Also, I find things like updating and/or installing packages to be confusing. Anytime I want to do something, it seems I need to install a package in which case I find the steps online. Once I execute the commands, I never understand what I've done, but know it works

who want to install packages from command line nowadays? recovery tools are now all in Windows GUI. in M$ Windows, you google, you download and click setup.exe, done! tell me what a Linux command line can do Windows (the GUI) cant? i stuck with AutoCAD just because of this. i got used to type quicker than finding which button on the ribbon bar in newer app such as Inventor or Solid Work etc (no command line) but i guess this is a curse from 1995 era. probably the same reason why people still use Eagle instead of KiCAD/Diptrace/Altium. i suggest you get rid of it while you still can, find an OS that can do everything on a Windows GUI, if you get used to them, things should be much quicker.

[Lindley]

Have to agree with @bostonman,  its not easy for the beginner or occasional user to understand where all these 'strange' commands come from.

Having used the Rasp Pi, when trying to set up some basic projects, you have to folllow web instructions for all the command line instructions to import things,  but theres no clear and easy way to understand how or why they are needed.

While its probalby easy in a work environment to pick up and discuss all these points from colleagues, the average home user alone can find it difficult to get to going with these details.

Any links to some good tutorials on these points most welcome.

[Nominal Animal]

Also, I find things like updating and/or installing packages to be confusing. Anytime I want to do something, it seems I need to install a package in which case I find the steps online. Once I execute the commands, I never understand what I've done, but know it works

who want to install packages from command line nowadays?

For Debian-derivatives, I can warmly recommend Synaptic, a GUI-based package manager interface.  It has very good search facilities (allowing you to search keywords in the package name, or package description, or in both).  It is not as powerful as the command-line interface, but for day-to-day tasks, it is very useful.  It also automatically tells you what other packages it marks to be installed if there are dependencies.

I do also use a couple of PPA's, or Personal Package Archives.  These are basically miniature repositories intended for individual software suites or families of applications.  Currently, I have the FreeCAD (stable) and OpenSCAD ones enabled, as the mainline repository versions are older than I like.  I recommend using the sudo apt-add-repository etc. commands shown for each repository to add that repository, run sudo apt update to reload repository package lists, and finally fire up Synaptic and look for the package to install.

The only software I have installed the Windows way -- via an installer instead of a repository and package management -- is Arduino + Teensyduino.  However, they install as the user running them (without superuser privileges), so that's okay.  (My udev rules for microcontroller support are customized, though.)  Although I really don't like the Arduino editor much, and have considered switching to e.g. PlatformIO.  One of the custom scripts I use with Arduino is one that reboots and autodetects Pro Micro clones' serial port (the cheap ATmega32u4's that have the Arduino Leonardo bootloader, and are treated as Arduino Leonardos in the Arduino environment, that you can get for ~ $5 USD apiece off eBay); I like those, but they can be a bit fiddly to program without that utility.  (It interposes avrdude, that's all.)

Like AVE says, the learning curve is steep, and you will suck at the beginning.  Some of that old DOS muscle memory may even be hindering you, because you already have expectations on how things should work, and you're finding out Linux works differently.  It is unfortunate, but I haven't found any way around it.  The sooner you'll decide it's just another tool that you wish to find how it works best for you, the easier it will be.
(I'm not saying you are not already doing that, I'm talking statistically, according to my own experience helping others learn.)

The Debian Reference Card (English PDF; others and Debian documentation at https://www.debian.org/doc/user-manuals) can be useful.

You can find the latest versions of the C library interfaces (sections 2 and 3) and command-line commands (sections 1, 6, and 8) at the Linux man pages project, which is the upstream for the non-application/library-specific man pages.  On the command line, you can use man section command-or-function to display a specific man page, say man 1 bash .  You can omit the section number, in which case man will look it up in the preferred section order, but note that sometimes different sections do have very different pages; for example, man 2 signal shows the signal() C library function interface, but man 7 signal shows the overview of POSIX/Unix signals.
You can do a keyword search using man -k term or man -s section -k term to look up manual pages related to term.

One very useful command in Debian derivatives is is dpkg-query -S $(which command) or equivalently dpkg-query -S /path/to/file .  It will tell you the package name that provided the command-line command or specified file; just remember to supply full absolute path to the file in the latter form.  Then, you can use dpkg-query -L package | less to list all the other files in that package, or dpkg-query -s package to show the description of the package.
If you don't remember those options (I don't!), use man dpkg-query or dpkg-query --help to look them up!

(This is also the reason why my example code almost always contains such an usage information when run with -h or --help as the first command line parameter.  I have a directory tree full of examples, and instead of checking the source code to see if a particular one is the one I'm looking for, I just run the binary with the --help flag, to see if it does what I'm looking for.  If it does, then I go looking at its sources.  Much faster this way.)


One thing I wish new Linuxers would consider and wield properly, is the privilege separation: users, groups (including supplementary groups), and filesystem capabilities.  If you ever find yourself doing sudo su - or chmod 0777 /dev/foo, you're doing things seriously wrong.

I have installed quite a few Apache servers in large organizations over the past two and a half decades, with multiple partially overlapping administrators (really, a proper admin hierarchy), without anyone needing to use sudo , with file user ownership indicating creator user, and access managed through group memberships (and a couple of helper scripts).  So I do claim I know this stuff.

On single-user workstations it does not matter much, except that it opens security holes that may or may not matter, but getting it right gives you a powerful new set of tools (that again, on a single-user workstation may not be of much use).  However, even a rough understanding the privilege separation mechanism gives you a starting point to solve problems instead of getting completely frustrated.  In short, that every process has a specific user and a group, plus optionally a set of supplementary groups (that are essentially equivalent to the process group); and optionally a set of Linux capabilities, with the superuser (root) having all capabilities; and files and devices are owned by a specific user and group; that udev sets those (and the access mode associated) for devices; and that executable binaries, but not scripts, can be set to grant special privileges (capabilities), much like the SetUID and SetGID bits can grant the process executing the binary superuser privileges, except much more fine-grained.

I wish there was a good guide I could point you to, but thus far, I haven't found one.
I have considered trying to write my own, several times, but the best format I can think of is a Wikipedia-like interlinked snippet archive, but it is too large an undertaking for myself alone, and there are very few people I'd trust to add correct advice...  and methinks providing misleading advice is much worse than being silent.

[tszaboo]

Well, on linux, if you press tab twice it lists out all the available commands.
And if you want more info on a command, you can type man <command> that should tell you what it does.
And yes, the operating system is a confusing mess the first, second and hundredth time you try to use it. Especially, that you have to interact not only with a bunch of built in commands, but distribution depended cli applications and setup files. The documentation of those is sometimes terrible, and the usual way of solving any problem on a linux computer is having a second computer where you goolge error messages.
So unless you really need to use Linux, why would you? There is the powershell on windows, all the commands you learned should be working there, it is familiar, and it is extended to have much more capabilities than the old cmd.exe. Give it a try.

[MIS42N]

There is no DOS vs. Linux, it's unfair to try and compare the two. It's like comparing a bicycle and a truck. They both have wheels, go on roads, but there are too many differences for meaningful comparison.

I have been running a Linux server for over 12 years. It has a mail server, web server, time server, DLNA, software RAID, DHCP, Samba and some other things and runs 24/7. I really don't know how it works. Setting up each component took many hours of reading and sorting out errors, and I learned that this (for me) had to be written down step by step because I didn't learn the commands, didn't know what each configuration did, didn't have a good grasp of the security, etc. But if I had to rebuild it from scratch, I think I could, using what I have written. My point is Linux can be used without understanding it. Every now and then it is a total pain because something goes wrong but it has happened to someone else so a bit of research usually finds the answer.

It would be different if Linux was my desktop. Then it would be very desirable to learn commands so I didn't have to refer to my notes for every little thing. I have an older laptop running Puppy Linux, mainly because I wanted a serial terminal to log output from microprocessors. But again I know just enough to do what I want.

As I see it, there are two paths. Treat Linux as a tool and learn the bits you need and treat the rest as a black box, or become a mechanic and learn what is under the hood. I have at times done some in depth learning but without practice it doesn't stick. So I now take the tool approach.

I should mention that Linux is a generic term for a great variety of operating systems that derive from a common root but can be very different. For the uninitiated it is best to stick to one with a huge user base, because most of the answers are out there for that flavour of Linux. Some learning transfers from one Linux to another, but often there is a variation. I came across Puppy Linux because I wanted something simple. It logs in automatically as the root user, which is anathema for a serious user but ideal for a novice like me that keeps nothing sensitive on it and don't want to deal with unnecessary complications.

When it comes to Linux, it is definitely "horses for courses".

[rstofer]

Mostly, we use the 'bash' shell.  Buy a book on using bash  Google 'bash shell book'  There are many
Sometimes we use the Bourne shell for scripting.  Buy a book on using sh
You may even run across the C shell.  Buy a book on using csh
Possibly even the Korn shell.  Buy a book on using ksh

The problem is that these books are not going to hold your hand learning the really simple commands.  So I Googled for 'raspian list of shell commands' - I have one of the new Pi 400 computers on my desk.  I got

https://www.raspberrypi.org/documentation/linux/usage/commands.md

This is a list of the most common commands.

There are hundreds of web sites that discuss shell commands.  You can absolutely forget learning all of them or even a significant percentage.  There are many commands and most have a bunch of options.  You really need a ready resource and I prefer books to web sites.  I'm probably a relic...

When you think you understand the options for 'tar', you don't.

The command line (shell) lays over the operating system and provides various utilities that allow the user to interact with the computer.  It's the same with CP/M, MSDOS and all the other operating systems.  But Unix/Linux takes command line functionality to a whole new level.

Too much typing?  You aren't using the 'alias' feature as much as you should.  You can type 'ls -la' or you can alias a new command 'la' to do the same thing.  Alias is one of my most favorite features.

[james_s]

who want to install packages from command line nowadays? recovery tools are now all in Windows GUI. in M$ Windows, you google, you download and click setup.exe, done! tell me what a Linux command line can do Windows (the GUI) cant? i stuck with AutoCAD just because of this. i got used to type quicker than finding which button on the ribbon bar in newer app such as Inventor or Solid Work etc (no command line) but i guess this is a curse from 1995 era. probably the same reason why people still use Eagle instead of KiCAD/Diptrace/Altium. i suggest you get rid of it while you still can, find an OS that can do everything on a Windows GUI, if you get used to them, things should be much quicker.

I do. The command line is so much faster and more efficient, and I can administer every machine in the house just by ssh into the terminal from my laptop. The GUI is easier for someone who is still learning their way around but the command line is superior when you know what you're doing.

[DiTBho]

On Unix there are no "commands".

"Commands" is a concept used in OS like DOS where everything you type is an applet handled by things like "command.com", which also offers a built-in shell, but it's not even close to the UNIX shell.

On Linux, UNIX and BSD, there are no "commands", there are only programs and two main big concepts: each program has three streams, and streams can be piped.

The three streams are:

• STDIN, namely input
• STDOUT, namely output
• STDERR, namely a second output channel usually use for signaling errors

The output of a program can be used to feed the input of a second program, and/or redirected to/from a file.

Code: [Select]

ls *.php | grep app
(silly example, but here there are to "programs" (ls, grep) piped: the output of "ls" feeds the input of "grep")

So, what you type on the bash belongs three categories

• internal bash variables and applets (e.g. "for", "loop", "if" ...), used to create a script
• operating system's environment, used to operate with scripts and programs
• the name/s of the program/s to be launched, plus options, pipe/s, and parameters

You can always type "man program_name" to see the full documentation online!

[james_s]

Semantics. From the end user standpoint there is no difference between a "command" and a "program". It just needlessly complicates things to even add that distinction for a beginner. Whether typing "ls" is "commanding" the computer to do something, or executing a program that does what the user is wanting to do is completely irrelevant.

[Karel]

Linux on the other hand is a OS made by geeks for geeks.

Linux is made by engineers, for engineers. It's true that it can attract geeks, but most users of Linux are engineers.

[golden_labels]

bostonman:
Where did the comparison between (MS-?)DOS and Linux(-based distribution) come from? Those operating system families has nothing to do with each other and are dissimilar on nearly any possible level. Perhaps this is where your confusion comes from?

Forget about (MS-?)DOS completely if you want to learn a Linux distro, some BSD or any other modern operating system. It’s like trying to learn drive a train engine by comparing it to a crank-started Ford Model T from 1910s.

If you do not feel comfortable with using shell directly, you may try some distribution that is designed to be more friendly towards people not wishing to learn by being thrown at the deep water with sharks around. This way you may gradually learn new things. Some will argue your progress will be slowed down and that is true, but I find it better to learn things in 2 years than giving up after a day of frustration and not learning at all.

You may also learn by managing some web hosting through SSH. That protects you from making Catastrophic Mistakes™, while you are still getting accustomed to using shell on daily basis. While there is little use for shell on Windows unless you are an admin or a programmer, you may use bash and nowadays Windows even has its own PowerShell.⁽¹⁾

As for software packages: see them as items in app stores, usually with an option to add other, custom stores (if you trust them). Instead of giving random people full, uncontrolled access to your computer, you ask a package manager to install stuff for you. How packages can affect your system is relatively well defined and mostly reversible from the very same tool.

A side note: avoid dd. Not only it has zero benefits for most operations, but its behavior is misunderstood and leads to unexpected outcomes. It’s also too easy to misuse, which led to its alternative name: disk destroyer. cat and head are more than enough, possibly supplemented with tee.

Why your dd command has failed is hard to tell, because I have no details. Most likely cause is the same as always, in any system: you made a wrong choice. Depending on the environment it may be wrong tools or options, or it may be an error made by an author of the program you use — in which case your mistake is trusting them . There is correlation between the platform used, but do not overestimate it. I have seen a ton of destruction by people using tools improperly on Windows, as well as horrible programs written for Linux. There even are whole distributions that make more advanced users tear hair from their heads.

____
⁽¹⁾ Though idea behind it is a bit different from that of shells on Unix-ish systems.

[james_s]

DOS and Linux command line are not THAT different on the surface, more like comparing a moped to a customized racing motorcycle. All of the common commands in DOS have Unix equivalents that do roughly the same thing. Most of the typical stuff like copying/moving/renaming files, launching executables, creating text files, installing software, etc can be done from either one using similar commands. Copy = cp, move = move, md = mkdir, rd = rmdir, cd = cd, etc.

[grumpydoc]

I confess I am wondering why we are talking about two dinosaurs - DOS vs Linux shells - in 2021? Didn't someone invent the GUI already - I hear even Linux systems have them these days 

[james_s]

I confess I am wondering why we are talking about two dinosaurs - DOS vs Linux shells - in 2021? Didn't someone invent the GUI already - I hear even Linux systems have them these days 

Of course they do, but the command line is still king. Once you know your way around it is MUCH faster to work in the command line. Many of my Linux systems are headless, there is no monitor or mouse connected to them and they are acessed remotely via ssh. Yes I could use VNC and have a graphical desktop but why? What benefit does that bring other than being slower, more tedious to use, and requiring more effort to set up? When I want to move files around and do other stuff of that nature I use the terminal, even on my Mac. It's probably an order of magnitude faster than opening finder, drilling down to the file I want, opening another finder window, drilling down to the place I want it, dragging the file over, then renaming it to what I want it called, or whatever else I'm trying to do. Instead I can type out one single command line that does it all.

[Mechatrommer]

Linux on the other hand is a OS made by geeks for geeks.

Linux is made by engineers, for engineers. It's true that it can attract geeks, but most users of Linux are engineers.

what is your basis of saying that?

i checked pcb market... https://smtnet.com/news/index.cfm?fuseaction=view_news&news_id=9591&company_id=53231
i checked cad market... https://www.cadalyst.com/cad/pandemic-transforms-cad-industry-2020-and-beyond-76189

i dont see any KiCAD nor Linux specific/specialized EDA/CAD taking a big chunk in the stats.. all point to softwares in Windows. are you saying..

1) engineers using Wine/VMware in Linux just to run few professional and expensive Windows softwares?
2) engineers/scientists that work from home, government or non profitable agencies? that never see the light in those statistics?

or am i missing something? show me some facts please.

[james_s]

It depends on your definition of "engineer". The FPGA toolchains at least for Altera and Xilinx are native *nix with Windows ports. Most of the big tech companies other than Microsoft use Linux internally, companies like Google employ thousands of software engineers who develop on Linux. "Geek" and "engineer" have a great deal of overlap.

[Mechatrommer]

and how many software engineers vs electrical/mechanical/civil/aeronautics/marine/chemistry/etc/etc/etc engineers in numbers out there?

Of course they do, but the command line is still king.

did you read my AutoCAD story earlier? we are both stuck. but i believe i'm not so serious as you are.

[james_s]

and how many software engineers vs electrical/mechanical/civil/aeronautics/marine/chemistry/etc/etc/etc engineers in numbers out there?

No idea, but I'm sure the data is out there.

Personally I know more software engineers than all other engineering disciplines combined, but I live in close proximity to Microsoft, Amazon, Google, and countless smaller tech campuses.

[grumpydoc]

I confess I am wondering why we are talking about two dinosaurs - DOS vs Linux shells - in 2021? Didn't someone invent the GUI already - I hear even Linux systems have them these days 

Of course they do, but the command line is still king. Once you know your way around it is MUCH faster to work in the command line. Many of my Linux systems are headless, there is no monitor or mouse connected to them and they are acessed remotely via ssh. Yes I could use VNC and have a graphical desktop but why? What benefit does that bring other than being slower, more tedious to use, and requiring more effort to set up? When I want to move files around and do other stuff of that nature I use the terminal, even on my Mac. It's probably an order of magnitude faster than opening finder, drilling down to the file I want, opening another finder window, drilling down to the place I want it, dragging the file over, then renaming it to what I want it called, or whatever else I'm trying to do. Instead I can type out one single command line that does it all.

Oh, don't get me wrong - I quite agree and, having been using Unix systems since the 1980's, spend a lot of my time typing at bash although I often do prefer to use VNC to remote systems for one reason or another,  not infrequently that it is useful to be able to run a browser on them - eg the headless Centos box (a PC-Engines APU) that I use as a router has the only physical Ethernet connection to the modem (for pppoe), the modem in turn offers no command line being one of the very few bits of network kit that is not running OpenWRT but it does have a web interface - so the only way I can access that is to have VNC on the router.

But the new-user introduction to Linux, today, shouldn't really need the command line at all; unless bostonman still boots his 10th gen i9 into DOS, or just uses Windows as an way to have more than one DOS command shell open at once, of course